UCSF

UC San Francisco Electronic Theses and Dissertations

Title
Anatomical studies of brainstem-evoked antinociceptive controls

Permalink
https://escholarship.org/uc/item/1x19r907

Author
Reichling, David B.

Publication Date
1989

Peer reviewed|Thesis/dissertation

eScholarship.org Powered by the California Digital Library

University of California
ANATOMICAL STUDIES OF BRAINSTEM-EVOKED ANTINOCICEPTIVE CONTROLS

by

DAVID B. REICHLING

DISSERTATION

Submitted in partial satisfaction of the requirements for the degree of

DOCTOR OF PHILOSOPHY

in

NEUROSCIENCE

in the

GRADUATE DIVISION

of the

UNIVERSITY OF CALIFORNIA

San Francisco

Committee in Charge

Deposited in the Library, University of California, San Francisco

 ii

copyright 1989
by
David B. Reichling
 iii

Dedication

This thesis is dedicated to Vandana, for everything; and to my mother,

who has contributed to all my work from its very conception.
 iv

Acknowledge ments

Dr. Allan Basbaum has been my advisor and teacher, generously

sharing with me his knowledge, experience, enthusiasm, wit, and wine. It is
my privilege to say that my association with Allan will probably always be the
single greatest influence on my career in science, and I sincerely thank him
for the experience. I would also like to thank all the past and present members
of the Pain Research Group at UCSF, who, under the leadership of Allan
Basbaum, Howard Fields, Peter Ralston, and Jon Levine, have created an

exciting research atmosphere, full of ideas. Thanks, also, to Peter Ohara and

Mary Heinricher for long discussions about science and camping, and also for
warning me. Finally, I would like to thank Allison Gannon, Bonnie Lord, and
Margaret Mayes for their technical, and friendly, assistance.
 Anatomical Studies of Brainstem-Evoked Antinociceptive Controls
David B. Reichling

ABSTRACT

Although the midbrain periaqueductal gray matter (PAG) is a focal

point of analgesia research, its internal circuitry has not been characterized.
This research project used anatomical techniques to study the neuronal
circuitry that mediates antinociceptive effects elicited from the PAG. Since it
has been suggested that tonically active GABAergic interneurons are
interposed between opioid elements in the PAG and the neurons that project to
the medullary nucleus raphe magnus (NRM), a series of experiments were
performed to characterize the relationship between GABAergic neuronal
elements and PAG-NRM projection neurons. First, the distributions of relevant
transmitters (GABA, B-endorphin, enkephalin, and dynorphin) were mapped
by light-microscopic immunocytochemistry. These maps showed that the
transmitters were confined to distinct regions of the PAG and that the
predicted GABA/opioid interactions were likely to occur in the ventrolateral
PAG. Second, combined electron-microscopic immunocytochemistry and
retrograde tracing established the neural substrate for a GABAergic control.
Approximately 50% of synaptic contacts onto PAG neurons retrogradely
labeled from the NRM were GABA-immunoreactive. A related electron

microscopic immunocytochemistry double-labeling study demonstrated that

serotonin-immunoreactive neurons in the dorsal raphe nucleus also receive

GABA-immunoreactive synaptic contacts. Third, it was determined that

although a significant proportion of projection neurons in the medulla are

GABA-immunoreactive, few GABA-immunoreactive neurons in the PAG project
to the NRM. Finally, a series of neuronal tract tracing studies were performed
 vi

to characterize the relationship between ascending and descending collaterals

of PAG neurons. Many PAG neurons retrogradely labeled from the NRM were
double-labeled by retrograde tracer injections into medial thalamic nuclei or
ventrobasal hypothalamus. In contrast, the ventrobasal thalamus, lateral

hypothalamus, amygdala, and cerebral cortex receive afferent projections

from the dorsal raphe nucleus, but not from the PAG. Together these studies
have provided information on the characteristics of the PAG-NRM projection
and on the nature of GABAergic controls that influence its activity.

**—
Signature of Advisor
 vii

CONTENTS

Title Page ----

Copyright Page ii

Dedication ................................................................................................................ iii

Acknowledgements ............................................................................................... iv

Abstract ........

Table of Contents - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
vii

Chapter I: Introduction .......................................................................................

Chapter II: B-endorphin, enkephalin, dynorphin, and GABA

immunoreactivity in the rat midbrain periaqueductal
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Chapter III: Axon Collaterals to the Hypothalamus and Thalamus

from Periaqueductal Gray Neurons that Project to Raphe
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Chapter IV: The Contribution of Brainstem GABAergic Circuitry to

Descending Antinociceptive Controls.
I. GABA-Immunoreactive Projection Neurons.............................
Chapter V: The Contribution of Brainstem GABAergic Circuitry to
Descending Antinociceptive Controls.
II. Electron Microscopic Immunocytochemical
Evidence of GABAergic Control over the Projection to
the Nucleus Raphe Magnus in the Rat...........................................
Chapter VI: Electron Microscopic Immunocytochemical Evidence for
Synaptic Interactions between GABAergic Terminals
and Serotonergic Neurons in the Dorsal Raphe Nucleus
of the Rat.….....…...........…...........................
INTRODUCTION

The midbrain periaqueductal gray matter (PAG) has been implicated in

a wide range of behaviors and physiological functions including
antinociception and analgesia' (Basbaum and Fields 1984; Fields and Basbaum
1978; Mayer and Price 1976), pain (Mehler 1969; Nashold C 1969; Valenstein
1965), lordosis (Sakuma and Pfaff 1979), vocalization (Jurgens and Pratt 1979;
Kanai and Wang 1962; Yajima et al. 1980), cardiovascular regulation
(McDougall et al. 1985), regulation of hormone secretion (Aulsebrook and
Holland 1969), feeding (Jenck et al. 1987; Skultety and Gary 1962; Waldbillig
1975), gut motility (Skultety 1959), thermoregulation (Tseng et al. 1980; Shen et
al. 1986) rage reactions (Hunsperger 1956; Kiser and Lebovitz 1956), fear,
anxiety (Nashold et al. 1969). This abundance of putative functions has
attracted investigators from a number of fields to study the PAG. Electrical
stimulation is commonly used to study PAG function, and for the most part, this
approach treats the PAG as a "black circuit box" from which efferents to
numerous areas may be activated. Thus, despite interest in the functions of
PAG projections to other parts of the brain, very little is known about the
neuronal circuitry within the PAG itself that controls activity in its efferent
pathways. The relative paucity of data on PAG internal circuitry probably
derives, in part, from features inherent to the PAG that do not lend it to study
by methods that have been applied successfully to other areas of the brain. In

1 The term "analgesia" means a reduction in the perception of pain. Animal

studies estimate analgesia by measuring a reduced response to noxious stimuli.
This effect is properly referred to as "antinociception." Although the
occurrence of antinociception and analgesia are often correlated, the
phenomena (and the terms) are not identical. I will use terms like
"stimulation-produced analgesia" that have come into general use, advisedly,
without intending to imply true analgesia.
 the following pages I describe some of these factors, and outline strategies that
the present research program has employed to begin characterizing the
antinociception-related circuitry of the PAG.
Anatomical Heterogeneity of PAG
Predictable spatial organization of neuronal elements in a region can
greatly facilitate analysis of neuronal circuits. For example, our present
understanding of circuitry in the cerebellum, cerebral cortex, and spinal cord
is founded on knowledge of the laminar organization of recognizable classes of
neurons within these regions. In contrast, the morphology of neurons in the
PAG ranges from small/fusiform to large/multipolar and these neurons are
not clustered into obvious morphological classes. Furthermore, PAG neurons
are arranged in a reticular fashion -- not organized into readily apparent
anatomical subdivisions. In fact, historically, atlases of the brain have either
represented the PAG without subdivisions (Berman 1968; Brown 1943; Cajal
1909) or have divided it into dorsal and ventral halves on embryological
grounds (Castaldi 1923), or into quadrants (Konig and Klippel 1963; Gillilan
1943). In fact, some recent studies that were specifically designed to identify
cytoarchitectural features in the PAG, did not detect cytoarchitectural
subdivisions, although a gradual increase in cell density, soma and dendrite
size, and degree of myelinization from the aqueduct outward was reported
(Mantyh 1982; Gioia et al. 1984).
In contrast to the view that the PAG cannot be regionated, some studies
report the existence of cytoarchitectural subdivisions. The subdivision

schemes are not identical, but they propose a similar combination of

concentric and radially oriented boundaries. Hamilton (1973) mapped the
distribution of three classes of neurons in Nissl-stained cat PAG, and detected

three major subdivisions. The nucleus medialis is an annulus around the

aqueduct containing sparse, small neurons. This region was recognized in
human by Olszewski and Baxter (1954), and later in rat by Beitz (1985a,b). In
addition, Hamilton's nucleus medialis extends in a band outwards to the

ventrolateral border of the PAG. Another subdivision, located directly dorsal

to the aqueduct, is also commonly recognized (Beitz 1985a,b; Castaldi 1923;
Hamilton 1973; Oleszewski and Baxter 1954). This subdivision contains densely
packed, medium-sized neurons. The lateral region of the PAG contains densely
packed large neurons and may be divided into dorsal and ventral parts (Beitz
1985a,b; Oleszewski and Baxter 1954). The pattern of cytochrome oxidase in the
PAG seems to agree with this type of subdivision scheme (Conti et al. 1988).
Although together, these studies argue convincingly that the PAG contains at
least three major regions, the lack of agreement on exact boundaries, even
among these very detailed studies, makes it clear that natural boundaries do
not exist in the PAG. Thus, the cytoachitectural approach to parcellating the
PAG has probably been played out nearly to its limit.
As an alternative to cytoarchitectural mapping of morphological classes
of neurons, the immunocytochemical approach can generate maps of
transmitter distributions; these maps may have more direct functional
significance. The earliest recognized, and most carefully mapped
histochemical locus within the PAG is the dorsal raphe nucleus (DRN;
Dahlström and Fuxe 1964). The DRN consists of a dense cluster of serotonergic
neurons located in the ventral PAG; it extends from the most caudal aspect of
the PAG to the level of the third nerve nucleus. Since these serotonergic
neurons are physiologically distinctive (Aghajanian et al. 1978), and their
axons project much more widely throughout the brain and spinal cord than do
axons of PAG neurons (Chapter III), the DRN is properly considered a nucleus
separate from, but embedded within the PAG. Therefore, my use of the term
"PAG" excludes the DRN.

Differential distributions of other transmitters in the PAG have been

noted by immunocytochemical surveys. Briefly, somatostatin (SS)

immunoreactive fibers are dense in the medial and ventrolateral PAG, in a

pattern similar to the peptides mentioned above, and a few SS cell bodies are on
the dorsal midline (Spangler and Morley 1987, cat; Krisch 1978, rat).
Adrenocorticotrophic hormone (ACTH) and FMRF-amide fibers, which
originate in the mediobasal hypothalamus, are also found in the medial and
ventrolateral PAG (Romagnano and Joseph 1983; Triepel and Grimmelikhuijzen
1984). Substance P immunoreactive neurons and fibers are concentrated

lateral to, but near the aqueduct; this pattern shifts dorsolaterally and dorsally
in the rostral PAG (Gioia et al. 1988; Moss and Basbaum 1983). Vasoactive

intestinal polypeptide (VIP) immunore active neurons are found only

subependymally, in the ventromedial PAG; sparse VIP immunoreactive fibers
are found in the ventral and ventrolateral PAG (Moss and Basbaum 1983, cat).

Cholecystokinin (CCK) immunoreactive neurons and fibers are found lateral to

the aqueduct at the level of the fourth nerve nucleus, and a dense group of
cells is in the Edinger-Westphal nucleus (Innis et al. 1979). 5

hydroxy tryptamine (5-HT) immunore active fibers are distributed evenly

throughout the PAG (Clements et al. 1985), as are histamine (histidine
decarboxylase immunore active) fibers, which derive from the posterior
hypothalamus (Watanabe et al. 1984). Glutamate and aspartate immunoreactive
fibers are spread evenly throughout the PAG. Although glutamate cells are
found throughout the PAG, aspartate cells are concentrated near the outer
edges of the rostral PAG (Clements et al. 1987). A few choline acetyltransferase
and tyrosine hydroxylase immunoreactive cell bodies are found in the
ventrocaudal PAG (Jones and Beaudet 1987). The results of these mapping
studies emphasize the fact that the PAG is composed of many superimposed
mosaics, which can never be represented by any single scheme of
intranuclear subdivisions.

Recently, five transmitters that have been implicated in

antinociception have been mapped in the PAG. These maps are particularly
relevant to the present series of studies. Shipley et al. (1987) produced a map
of the distribution of immunoreactive neurotensin in the rat, and Chapter II
reports our observations on the distribution of GABA, B - end orph in ,
enkephalin, and dynorphin in the rat. Immunoreactive staining for all five
transmitters is most intense in the region corresponding to Hamilton's (1973)
nucleus medialis. This pattern must, in part, reflect the greater density of
terminals versus cell bodies in this region, but since some transmitters
(described above) are not found in this pattern, the relationship to Hamilton's
scheme is probably significant. Beyond this overall likeness, however, each
pattern of staining has some prominent unique features (see discussion in
Chapter II). As described later in this Introduction and also in Chapter II,
these observations are useful in making predictions about antinociception
related neuronal interactions and their localization in the PAG.

Light microscopic observations, however, can be easily

overinterpreted. For example, it may not be safe to assume that
the distribution of transmitters in a region is a good indication of the
distribution of their postsynaptic targets. There is evidence that this

commonsense arrangement is sometimes violated in the brain (for review see

Herkenham 1987), and in this respect the PAG is a major trouble spot,
particularly in regard to transmitters that are of special interest to the present
series of studies. As Chapter II describes in detail, prominent "mismatches"
exist between the distribution of opioid peptides and their receptors in the
PAG. I have used light microscopic immunocytochemical maps to hypothesize
certain neuronal interactions and to suggest where conditions are permissive
for them to occur. Demonstration that neural connections, in fact, exist

requires the use of much more diffiucult techniques.

Electrophysiology
The anatomical heterogeneity of the PAG might result in variations in
the electrophysiological properties of neurons according to their locations. a
problem that must be confronted by such physiological studies is that PAG
neurons exhibit a wide variety of responses to peripheral stimuli, and it is not
clear how these responses should be divided into physiological classes. This
problem is compounded by the general lack of consesus over how PAG neurons
respond to peripheral stimuli or to opiates (for review see Gebhart 1982).
Using a simple means of classifying responses to heat stimuli, Sanders et al.
(1980) reported that most responsive neurons in the PAG are excitated, while
inhibitory responses predominate in the DRN. Schurr et al. (1981) reported
that neurons in the ventral PAG are distinctive in that they have much higher
spontaneous activity than other PAG neurons, in the morphine dependent rat.
Liebeskind and Mayer (1971) found evidence of somatotopy in PAG evoked
potentials. This somatotopy might be a subtle property distributed over the
population of PAG neurons, since single-unit recording studies do not indicate
somatotopic organization of the large receptive fields of PAG neurons.
One approach to this problem of defining physiological classes of
neurons in the PAG analyses the relationship between their activity and
nocifensive behaviors, instead of concentrating on somatosensory responses.
Many neurons in the midbrain change their firing rate immediately before
Such behavior (Handwerker and Sack 1982; Korzeniewska et al. 1986), and
 these neurons are most common in ventral than in dorsal PAG (Heinricher et

al. 1987). This approach is promising, but it is clear that our present
understanding of PAG electrophysiology is still sketchy.
Functional Heterogeneity -- Electrical Brain Stimulation
In view of the intermixture of anatomical and electrophysiological
characteristics in the PAG, it is optimistic to expect to find discrete localization
of functions within the PAG. In fact, functional studies of the PAG have not,
for the most part, found distinct zones, but instead reported gradual changes
across relatively large distances. For example, Skultety (1963) reported that
stimulation of rostral PAG produced rage reactions in cats, while stimulation at
caudal sites produced escape behavior. Most reports of functional
heterogeneity have described differences along the dorsoventral dimension.
In general, the electrical stimulation in the dorsal PAG is aversive (Cosyns and
Gybels 1979; Delgado 1955; Fardin et al. 1984b; Hunsperger 1956; Kelly et al
1946; Kiser and Lebovitz 1975; Magoun et al. 1937; Nashold et al. 1969; Olds and
Olds 1962; Valenstein 1965; Wada et al. 1970; Wolfle et al. 1971). On the other
hand, electrical stimulation in the ventral PAG is rewarding (Belluzzi and
Stein 1977; Schenberg and Graeff 1978; Schmitt et al 1974). Motor phenomena
are also associated mainly with ventral PAG stimulation (Blackburn et al. 1980;
Fardin et al. 1984a; Kiser et al. 1978; Nicolau et al. 1979; Skultety 1962). This
localization of reward and motor effects of stimulation in the ventral PAG may
reflect the presence of the DRN there -- serotonergic pathways from the DRN
have been implicated in both phenomena.
Numerous studies have mapped the PAG for sites that support
stimulation-produced analgesia (SPA). Soper and Melzack (1982) reported that
dorsal (and rostral) electrical stimulation produced antinociception against
pinch applied to the rostral body surface, while ventral (and caudal)
 stimulation was more effective for caudal stimuli. Generally, it is agreed that
although antinociception may be elicited from most parts of the PAG
(especially in rats), the most effective region is the ventral PAG (Oliveras et al.
1974; Yeung et al. 1977) More specifically, some studies have singled out a
region at the ventrolateral edge of the PAG (Dennis et al. 1980; Gebhart and
Toleikis 1978; Lewis and Gebhart 1977). In addition, the nature of

antinociception elicited from dorsal and ventral sites may differ. Liebeskind
and coworkers demonstrated that tail-flick analgesia elicited from ventral, but
not dorsal, PAG sites is reduced by naloxone pretreatment or NRM lesions
(Cannon et al. 1982; Prieto et al. 1983; however, see Klatt et al. 1988).
Furthermore, post-stimulation antinociception is more often elicited from
dorsal, than ventral, PAG (Cannon et al 1982; Fardin et al. 1984b,c). Fardin et
al. (1984b,c) also noted that antinociception can be elicited from the ventral
PAG unaccompanied by behavioral effects, while dorsal PAG analgesia was not
separable from aversive side effects. This linkage of aversion and
antinociception in the dorsal PAG suggested that antinociception from this
region might result indirectly from stress induced by the aversive effects of
stimulation. Contrary to this proposal is the observation by Morgan et al.
(1987) that systemic benzodiazepine can, in fact, dissociate the antinociceptive
and aversive effects of stimulation in the PAG.

Much emphasis has been placed on the discrete zones in the ventral
PAG where SPA can be elicited in the absence of any aversive side effects.

While the location of these "pure analgesia areas" has obvious clinical
importance, its relevance to the analysis of antinociceptive circuitry is
questionable. From the foregoing discussion of the intermingled systems in
the PAG, it seems very unlikely that any site in the PAG will be found
exclusively dedicated to a single physiological function. The relevance of
neuronal circuitry at a site should not be underestimated only because it is
intermingled (and perhaps interrelated) with other functional systems.
Since some minimum number of relevant neurons must be activated in

order to produce an observable effect by electrical stimulation, the technique

is inherently limited in its resolution. By exciting all neurons within this
volume of tissue, electrical stimulation overrides local synaptic circuitry,
exciting axons that project from, and through, the stimulated area. These

axons are uniformly excited (orthodromically and antidromically) in non

physiological patterns. In this way, electrical stimulation maps only the
distribution of projection neurons (and axons of passage) that mediate an
effect, and in this manner it is very nearly the inverse of making a lesion.
For these reasons, electrical stimulation is very poorly suited to task of
analyzing synaptic circuitry local to an area.
Functional Heterogeneity -- Microinjection
A more physiological means of activating PAG circuitry is the
microinjection of drugs and transmitters. Glutamate has often been

considered a non-selective excitant of neuronal cell bodies, and is used in lieu

of electrical stimulation to avoid stimulating axons of passage, or

antidromically activating other axons. Although glutamate microinjection in
the PAG causes potent antinociception (Behbehani and Fields 1978) and many
components of the defense reaction (for review see Bandler and Depaulis
1988), as yet this technique has not been used to map the region.
Interestingly, N-methyl-D-aspartate, a specific NMDA receptor ligand, also
causes antinociception when microinjected in the PAG (Jacquet 1988). Thus, it
is risky to assume that the antinociceptive effect of glutamate results from a
non-selective excitation of neurons in the PAG.
 1 0

Opiate microinjection is another more physiological means of activating

PAG circuitry. It is often used to elicit antinociception from the PAG, and some
regional differences have become apparent. In general, the most efficacious
sites for producing antinociception by opiate microinjection lie in the
ventrocaudal PAG (for review see Yaksh and Rudy 1976). More specifically,
Lewis and Gebhart (1977) found that microinjected morphine produced the
most potent antinociception when microinjected in the ventromedial PAG.
Similarly, Sharpe et al. (1974) reported that antinociception was elicited with
the shortest latency by morphine microinjected in the ventromedial PAG. The
greater efficacy of medial, compared to lateral, morphine injections might be
caused, in part, by spread into the ventricle, giving the drug faster access to a
greater area of PAG tissue.
Together, the physiological and functional studies discussed above
suggest that the major spatial variation of function in the PAG occurs between
its dorsal and ventral parts. As mentioned previously, these differences are
probably due, in part, to the prsence of the DRN in the ventral PAG. In
addition, however, neuronal tracing studies reveal that the dorsal and ventral
parts of the PAG have markedly different afferent connections with other
regions of the brain. The ventral PAG receives projections from limbic,
autonomic, and motor systems, while the dorsolateral PAG receives
predominantly sensory inputs (Beitz 1982). While this segregation of inputs
suggests that different functions are mediated by these regions, it also raises
suspicions that some of the apparent functional differences demonstrated with
electrical stimulation may be due entirely to antidromic activation of afferents
to the PAG.

Strategic Considerations
 1 1

Compared to the practical difficulties with PAG research, described

above, a more conceptual obstacle to progress is the poor understanding of
PAG functions. Although electrical stimulation of the PAG influences a wide
range of brain processes and associated behaviors, it is not clear how these
artifacts of stimulation are related to actual physiological functions of the PAG.
The results of electrical stimulation are particularly difficult to interpret for
an area that has afferent and efferent connections with so many other areas
of the brain. (Chapter III discusses some serious difficulties in interpreting
the results of electrical stimulation in the PAG, that were revealed by new
anatomical data.) As a result, most functional proposals concerning the PAG
have been formulated only in the most general terms, e.g. "control of feeding."
In fact, many cannot be considered proper functional descriptions, but rather
statements of suspected involvement of the PAG in some process or behavior.
This problem results, in part, from the very nature of these phenomena --
many are diffusely distributed processes in the brain (e.g. arousal), and the
individual participation of the PAG may not yield to simple verbal descriptions.
Further progress in sorting out and defining functions of the PAG will
require a more sophisticated understanding of the neuronal circuitry within
the PAG itself. One logical approach to the problem of characterizing local
neuronal circuitry is to first determine which efferents from the structure
mediate a particular behavior or physiological effect, and which inputs,
upstream in the local circuitry, are able to activate these efferents. From

knowledge of the input/output relationship between these two points, it should

be possible to make testable predictions about the interposed neuronal
circuitry. This strategy is most practical if the relevant afferent and efferent
pathways are well-characterized and accessible. Sorting out pathways to and
from the PAG, however, can be a diffiicult task because of the great number of
 1 2

interconnections between this and other regions of the brain.” Because of

these extensive interconnections between the PAG and other regions of the
brain, relevant afferent and efferent pathways are only poorly characterized
for most of the phenomena induced by electrical stimulation of the PAG.
Model System -- Lordosis
One putative function of the PAG that has been described in relatively
specific terms relates to reproductive behavior in female rats. Electrical
stimulation of the PAG markedly facilitates lordosis, a dorsiflexion reflex of
sexually receptivity that is elicited in estrogen-treated female rats by somatic
stimulation of the hindquarters (Sakuma and Pfaff 1979a). Conversely, lesions
of the PAG abolish the reflex (Sakuma and Pfaff 1979b). A series of

experiments by Pfaff and others have begun to outline the neuronal circuitry
involved in this phenomenon. Electrical stimulation of the ventromedial

nucleus (VMN) of the hypothalamus facilitates the lordosis reflex; this effect is
abolished by PAG lesions (Sakuma and Pfaff 1979b). Based on these data, it has

2 The PAG receives major afferent projections, in various species, from medial
prefrontal cerebral cortex, amygdala, diagonal band of Broca, hypothalamic
nuclei (dorsal premammillary, ventromedial, lateral, posterior, anterior,
perifornical, preoptic, and tubercinereum), cuneiform n., subcuneiform n.,
substantia nigra, ventral tegmental area, locus coeruleus, parabrachial area,
mesencephalic, pontine, and medullary reticular formation, n. raphe magnus,
pallidus, obscurus, and centralis, trigeminal n. and spinal cord (dorsal horn,
lateral cervical n.) (Beitz 1982; Grofová, et al. 1978; Liu 1983; Marchand and
Hagino 1983; Meller and Dennis 1986; Swett, et al. 1985; Wiberg, et al 1987).
The PAG sends major efferent projections, in various species, to lateral
habenula, zona incerta, thalamic nuclei (n. reticularis, mediodorsal n.,
laterodorsal n., lateral posterior n., midline group, and intralaminar group),
hypothalamic nuclei (preoptic n., lateral n., posterior n., anterior n., dorsal n.,
periventricular n., ventromedial n., periarcurate area), superior colliculus,
cuneiform n., dorsal tegmental n., locus coeruleus, lateral parabrachial n., n.
reticularis, pontine reticular n. oralis and caudalis, medullary gigantocellular
n., paragigantocellular n., n. raphe magnus and pallidus, n. ambiguus, (Abols
and Basbaum 1981; Beitz, et al. 1983; Chi 1970; Fardin, et al. 1984a; Gallager and
Pert 1978; Hamilton 1974; Hamilton and Skultety 1970; Mantyh 1983a,b; Ruda
1975).
 1 3

been proposed that the VMN initiates lordosis via its efferent connection with
the PAG. Since neurons in the VMN are sensitive to estrogen levels, this
pathway is probably responsible for the estrogen-sensitivity of the behavior.
A second necessary stimulus for lordosis is pressure or noxious stimuli to the
hindquarters. The PAG receives such input directly via spinopetal projections,
and indirectly from the reticular formation, and perhaps from the thalamus.
These inputs are presumably integrated in the PAG and activity in some
efferent pathway is accordingly modified. Evidence suggests that the relevant
efferent target of the PAG is the nucleus reticularis gigantocellularis (NGc) of
the medulla. Lesions in the lateral medullary reticular formation disrupt
lordosis behavior (Modianos and Pfaff 1979), and electrical stimulation of the

NGc monosynaptically excites motor neurons that innervate muscles of the

ipsilateral back which perform the dorsiflexion (Peterson et al. 1979).
Demonstrating links within the PAG that interconnect the afferent and
efferent pathways for lordosis has, however, been problematical. Consistent
with their model, Sakuma and Pfaff (1980a) showed that more than half the

units they recorded in the PAG responded both to electrical stimulation of the
VMN and to cutaneous noxious or pressure stimuli. The same investigators,
however, were unable to demonstrate somatosensory responses in any of 57
PAG neurons antidromically activated from the gigantocellular and
magnocellular nuclei of the medulla (Sakuma and Pfaff 1980b). As a result, it
was concluded that these data "preclude the involvement of these cells in the

[stimulus-bound] reflexive control of lordosis," and that the involvement of

the PAG in the reflex is probably limited to "transmit■ting] hormonal effects

on the forebrain and hypothalamus to the reticulospinal system."
Model System -- Antinociception
 1 4

Compared to lordosis, the phenomenon of antinociception may be more

amenable to the kind of input/output approach suggested above. The

hypothesis that the PAG is involved in an endogenous pain inhibitory system

has been reviewed extensively (Basbaum and Fields 1984; Fields and Basbaum
1978; Liebeskind et al. 1976; Mayer and Price 1976). Stated simply, the PAG is
thought to integrate information about internal and external states and, under
the appropriate conditions, it activates bulbospinal inhibitory pathways via
direct projections to the medullary reticular formation. Activation of these
pathways specifically inhibits noxious-evoked activity in spinal cord dorsal
horn neurons, including spinothalamic tract neurons. This apparent function
of the PAG has received far more attention than any other this antinociceptive
system here are several features of that make it a relatively convenient
paradigm for studying the neuronal circuitry of the PAG.
First, the system is activated by opiates microinjected in PAG.
Therefore, one important input to PAG antinociceptive circuitry can be
readily identified in anatomical studies (by immunocytochemistry), and
physiological studies can mimic and manipulate this input using a wide range
of specific ligands. Three classes of opioid peptides and receptors are present
in the PAG (see discussion, Chapter II). It is also fortuitous that 3-end orphin
terminals derive exclusively from the hypothalamic arcuate nucleus, and so
this input to the PAG is accessible to stimulation or lesions. In fact,

experiments in which the arcuate nucleus was lesioned have implicated 3

endorphin in analgesia produced by stimulation in the PAG (Millan et al.
1986).
Secondly, it is well-established that the projection from the PAG to the
nucleus raphe magnus (NRM) of the rostral ventromedial medulla mediates, in
part, antinociceptive effects elicited from the PAG (see Basbaum and Fields
 1 5

1984). This is the single largest descending projection from the PAG, and so
using anatomical or electrophysiological methods it is relatively easy to
identify neurons from which it originates. This pathway is not, however,
homogeneous. It contains a mixture of transmitters: approximately 32% of
neurons retrogradely labeled from the NRM contain neurotensin, 13%
somatostatin, 9% serotonin, and 2% GABA (Beitz et al. 1983; Chapter III), and
nearly half the transmitter content of the pathway is still unaccounted for.
The neuronal tracing experiments described in Chapter III reveal an
interesting anatomical heterogeneity in the pathway. A large proportion of
PAG neurons at the origin of the projection to the NRM also send collaterals to
the ventrobasal hypothalamus and to the medial thalamus. Thus, activation of
this pathway necessarily activates projections to other areas of the brain that
have been implicated in nociception and antinociception.
Circuit Hypotheses
Although relevant inputs to, and outputs from, the PAG have been
identified, very little is known about the neuronal circuitry within the PAG
through which antinociceptive effects are generated. Based on available

information, it is possible to make specific, testable hypotheses. Since

electrical stimulation of, or glutamate microinjection into, the PAG does

produce analgesia (Behbehani and Fields 1978), and since electrolytic lesion of
the PAG does not (Deakin and Dostrovsky 1977), it follows that PAG neurons
that project to the NRM must be excited in order to produce analgesia.
Opioidergic synapses, however, are believed to be exclusively inhibitory
(Nicoll and Alger 1979). Thus, for opiates to excite the PAG output neuron, it
has been proposed that opioid peptides inhibit an interneuron that tonically
inhibits the output neuron (Yaksh et al. 1976). In other words, opiates
disinhibit the PAG output neuron. GABA has been suggested as a likely
 1 6

transmitter for the inhibitory interneuron (Basbaum and Fields 1984). As

discussed in Chapter V, numerous pharmacological, physiological, and

anatomical observations are consistent with this proposal.
Recent technical developments in electron microscopic immunocyto
chemical and tracing techniques have made it possible to begin verifying that
the anatomical substrate for this hypothetical circuitry, in fact, exists. The
experiments in Chapter V were designed to establish that there are GABA
immunoreactive synaptic contacts onto some PAG neurons that project to the
NRM. The latter were identified by injection of the colloidal gold labeled
tracer, WGA-apoRRP-Au, into the NRM. Putative GABAergic terminals were
identified using an antibody directed against glutaraldehyde-fixed GABA.
These experiments not only demonstrated the existence of this synapse, but
also characterized the morphology of GABAergic synapses in the PAG, in
general.
The DRN has also been implicated in the antinociceptive effect of
opiates, and there is evidence that the serotonergic projection neurons in this
nucleus mediate part of the antinociception induced by opiates. In addition,
interactions between GABA and serotonergic neurons in the DRN have been
implicated in cardiovascular regulation and in the anxiolytic action of
benzodiazopines (see discussion, Chapter VI). Therefore, I also used

electronmicroscopic immunocytochemical techniques to verify that GABA

immunoreactive axon terminals are in synaptic contact with serotonin
immunoreactive cell bodies in the DRN. These experiments are described in
Chapter VI.
Conclusions

This series of experiments have contributed toward beginning the

characterization of antinociception-related circuitry in the PAG by:
 1 7

1) providing detailed information about the topography of antinociception

relevant transmitters in the PAG (Chapter II);
2) characterizing the NRM-PAG projection pathway by:
b) describing major collateral pathways deriving from the PAG-NRM
projection, that are likely to mediate effects related to antinociception
(Chapter III), and
a) investigating the contribution of GABAergic projection neurons to the
pathway (Chapter IV);
3) providing evidence for GABAergic control over the PAG-NRM projection,
and also over the serotonergic neurons of the DRN (Chapters V and VI);
These experiments are obviously just a beginning in the effort to characterize
the circuitry in the PAG that mediates antinociception. The results leave many
questions unanswered, but also suggest a number of new experiments to help
move our understanding of PAG circuitry further along upstream.
 18

REFERENCES

Aghajanian, G.K., Wang, R.Y. and J.M. Baraban (1978) Serotonergic and non
serotonergic neurons of the dorsal raphe: reciprocal changes in firing
induced by peripheral nerve stimulation. Brain Res. 153: 169-175.
Aulsebrook, L.H. and R.C. Holland (1965) Central inhibition of oxytocin release.
Am. J. Physiol. 216:830-841.
Bandler, R. and A Depaulis (1988) Elicitation of intraspecific defence reactions
in the rat from midbrain periaqueductal grey by microinjection of kainic
acid, without neurotoxic effects.

Basbaum, A.I. and H.L. Fields (1984) Endogenous pain control systems:
brainstem spinal pathways and endorphin circuitry, Ann. Rev. Neurosci.
7: 309-338.

Behbehani, M.M. and H.L. Fields (1978) Evidence that an excitatory connection
between the periaqueductal gray and nucleus raphe magnus mediates
stimulation-produced analgesia. Brain Res. 149: 266-269.
Beitz, A.J. (1982) The organization of afferent projections to the midbrain
periaqueductal gray of the rat. Neurosci. 7: 133-159.
Beitz, A.J., Shepard, R.D. and W.E. Wells (1983) The periaqueductal gray-raphe
magnus projection contains somatostatin, neurotensin and serotonin but
not cholecystokinin. Brain Res. 261: 132-137.
Beitz, A.J. (1985a) The midbrain periaqueductal gray in the rat. I. Nuclear
volume, cell density, orientation, and regional subdivisions, J. Comp.
Neurol. 237: 445-459.

Beitz, A.J. (1985b) The midbrain periaqueductal gray in the rat. II. A Golgi
analysis. J. Comp. Neurol. 237: 460-475.
Belluzzi, J.D. and L. Stein (1977) Enkephalin may mediate euphoria and drive
reduction reward. Nature 226: 556-558.

Berman, A.L. (1968) The Brainstem of the Cat. A Cytoarchitectonic Atlas with
Stereotaxic Coordinates. University of Wisconsin Press, Madison.
Blackburn, T.P., Foster, G.A., Heapy, C.G. and J.D. Kemp (1980) Unilateral 5,7-
dihydroxytryptamine lesions of the dorsal raphé nucleus (DRN) and rat
rotational behavior. Euro. J. Pharmacol. 67: 427-438.

Brown, J.O. (1943) The nuclear pattern of the non-tectal portions of the
midbrain and isthmus in the dog and cat. J. Comp. Neurol. 78: 365-406.
Cajal, Ramón y, S. (1909) Histologie du Systeme Nerveux, Vol. 1., Consejo
superior de Investigaciones Cientificas, Madrid.
 1 9

Cannon, J.T., Prieto, G.J. and Liebeskind, J.C., Evidence for non-opioid forms of
stimulation-produced analgesia in the rat, Brain Res., 243 (1982) 315-321.
Carrive, P., Dampney, R.A.L. and R. Bandler (1987) Excitation of neurones in a
restricted portion of the midbrain periaqueductal grey elicits both
behavioural and cardiovascular components of the defence reaction in the
unanesthetised decerebrate cat. Neurosci. Lett. 81: 273-278.

Castaldi, L. (1923) Studi sulla strutture e sullo svillupo del mesencephalo I.

Ricerche in Cavia cobaya. Arch. Ital. Anat. e Embr. 20: 23-225.
Chi, C.C. (1970) An experimental silver study of the ascending projections of
the central gray substance and adjacent tegmentum in the rat with
observations in the cat. J. Comp. Neurol. 139: 259-272.
Clements, J.R., Beitz, A.J., Fletcher, T.F. and M.A. Mullet (1985)
Immunocytochemical localization of serotonin in the rat periaqueductal
gray: a quantitative light and electron microscopic study. J. Comp. Neurol.
236: 60-70.

Clements, J.R., Madl, J.E., Johnson, R.L., Larson, A.A. and A.J. Beitz (1987)
Localization of glutamate, glutaminase, as partate and as partate
aminotransferase in the rat midbrain periaqueductal gray. Exp. Brain Res.
67: 594-602.

Conti, F., Barbaresi, P. and M. Fabri (1988) Cytochrome oxidase histochemistry

reveals regional subdivisions in the rat periaqueductal gray matter.
Neurosci. 24: 629-633.

Cosyns, P. and J. Gybels (1969) Electrical central gray stimulation for pain in
man. In J.J. Bonica, D. Albe-Fessard and J.C. Liebeskind (Eds.), Advances in
Pain Research and Therapy, Vol. 3, Raven Press, New York, pp. 511-5.14.
Delgado, J.M.R. (1955) Cerebral structures involved in transmission and
elaboration of noxious stimulation. J. Neurophysiol. 18: 261-275.
Dennis, S.G., Choinière, M. and R. Melzack (1980) Stimulation-produced
analgesia in rats: assessment by two pain tests and correlation with self
stimulation. Exp. Neurol. 68: 295-309.
Dahlström, A. and K. Fuxe (1964) Evidence for the existence of monoamine
containing neurons in the central nervous system. I. Demonstration of
monoamines in the cell bodies of brainstem neurons. Acta Physiol. Scand.
62: 1-55.

Dostrovsky, J.O. and J.F.W. Deakin (1977) Periaqueductal gray lesions reduce
morphine analgesia in the rat. Neurosci. Lett. 4: 99-103.
Fardin, V., Oliveras, J.L. and J.M. Besson (1984a) Projections from the
periaqueductyal gray matter to the B3 cellular area (nucleus raphe
magnus and nucleus reticularis paragigantocellularis) as revealed by the
retrograde transport of horseradish peroxidase in the rat. J. Comp. Neurol.
223: 483-500.
 20

Fardin, V., Oliveras, J.L., and J.M. Besson (1984b) A reinvestigation of the
analgesic effects induced by stimulation of the periaqueductal gray matter
in the rat. I. The production of behavioral side effects together with
analgesia. Brain Res. 306: 105-123.
Fardin, V., Oliveras, J.L., and J.M. Besson (1984c) A reinvestigation of the
analgesic effects induced by stimulation of the periaqueductal gray matter
in the rat. II. Differential characteristics of the analgesia induced by
ventral and dorsal PAG Stimulation. Brain Res. 306 125-139.

Fields, H.L. and A.I. Basbaum (1978) Brain stem control of spinal pain
transmission neurons. Ann. Rev. Physiol. 40: 193-221.
Gallager, D.W. and A. Pert (1978) Afferents to brain stem nuclei (brain stem
raphe, nucleus reticularis pontis caudalis and nucleus gigantocellularis)
in the rat as demonstrated by microiontophoretically applied horseradish
peroxidase. Brain Res. 144; 257-275.
Gebhart, G.F. (1982) Opiate and opioid peptide effects on brain stem neurons:
relevance to nociception and antinociceptive mechanisms. Pain 12:93-140.
Gebhart, G.F. and J.R. Toleikis (1978) An evaluation of stimulation-produced
analgesia in the cat. Exp. Neurol. 62: 570-579.
Gillilan, L.A. (1943) The nuclear pattern of the nontectal portion of the
midbrain and isthmus in rodents. J. Comp. Neurol. 78; 213-251.
Gioia, M., Bianchi, R. and G. Tredici (1984) Cytoarchitecture of the
periaqueductal gray matter in the cat: a quantitative Nissl study. Acta Anat.
119: 113-117.

Gioia, M. and R. Bianchi (1988) The distribution of substance P and met

enkephalin in the periaqueductal gray matter of the rat. Bas. Appl.
Histochem. 32: 103-108.

Grofová, I., Ottersen, O.P. and E. Rinvik (1978) Mesencephalic and diencephalic
afferents to the superior colliculus and periaqueductal gray substance
demonstrated by retrograde axonal transport of horseradish peroxidase in
the cat. Brain Res. 146: 205-220.

Hamilton, B.L. (1973) Cytoarchitectural subdivisions of the periaqueductal gray

matter in the cat. J. Comp. Neurol. 149: 1-28.
Hamilton. B.L. and M.F. Skultety (1970) Efferent connections of the
periaqueductal gray matter in the cat. J. Comp. Neurol. 139: 105-114.
Hamilton, B.L. (1974) Projections of the nuclei of the periaqueductal gray
matter in the cat. J. Comp. Neurol. 152: 45-58.
Handwerker, H.O. and R. Sack (1982) Single unit activity in the rat's midbrain
during nocifensive tail flick reaction. Neurosci. Lett. 30: 79-84.
 2 1

Heinricher, M.M., Cheng, Z.F. and H.L. Fields (1987) Evidence for two classes of
nociceptive modulating neurons in the periaqueductal gray. J. Neurosci. 7:
271-278.

Herkenham, M. (1987) Mismatches between neurotransmitter and receptor

localizations in brain: observations and implications. Neurosci. 23: 1-38.
Hunsperger, R. (1956) Role of substantia grisea centralis mesencephali in
electrically induced rage reactions. Prog. Neurobiol. 1:289-294.
Innis, R.B., Correa, F.M.A., Uhl, G.R., Schneider, B. and S.H. Snyder (1979)
Chole cytokin in octapeptide-like immuno reactivity: histochemical
localization in rat brain. Proc. Natl. Acad. Sci. 76: 521-525.

Jacquet, Y.F. (1988) The NMDA receptor: central role in pain inhibition in rat
periaqueductal gray. Euro. J. Pharmacol. 154: 271-276.
Jenck, F., Quirion, R. and Wise R.A. (1987) Opioid receptor subtypes associated
with ventral tegmental and periaqueductal gray inhibition of feeding.
Brain Res. 423: 39-44.

Jones, B.E. and A. Beaudet (1987) Distribution of acetylcholine and

catecholamine neurons in the cat brainstem: a choline acetyltransferase
and tyrosine hydroxylase immunohistochemical study. J. Comp. Neurol.
261: 15-32.

Jurgens, U. and R. Pratt (1979) Role of the periaqueductal grey in vocal

expression of emotion. Brain Res. 167: 367-378.
Kanai, T. and S.C. Wang (1962) Localization of the central vocalization
mechanism in the brainstem of the cat. Exp. Neurol. 6: 426-434.
Kelly, A.H., Beaton, L.E. and H.W. Magoun (1946) A midbrain mechanism for
facio vocal activity. J. Neurophysiol. 1: 181-189.
Kiser, R.S. and R.M. Lebovitz (1956) Role of substantia grisea centralis
mesencephali in electrically-induced rage reactions. In J. Ariens Kapper
(Ed.) Progress in Neurobiology, Elsevier, New York, pp. 289-292.
Kiser, R.S. and R.M. Lebovitz (1975) Monoaminergic mechanisms in aversive
brain stimulation. Physiol. Behav. 15: 47-53.
Kiser, R.S., Lebovitz, R.M. and D.C. German (1978) Pharmacologic and anatomic
differences between two types of aversive midbrain stimulation. Brain
Res. 155: 331-342.

Klatt, D.S., Guinan, M.J., Culhane, E.S., Carstens, E. and L.R. Watkins (1988) The
dorsal raphe nucleus: a re-evaluation of its proposed role in opiate
analgesia systems. Brain Res. 447: 246-252.
König, J.F.R. and R.A. Klippel (1963) The Rat Brain, A Stereotaxic Atlas of the
Forebrain and Lower Parts of the Brain Stem, Williams and Wilkins,
Baltimore.
 22

Korzeniewska, E., Brinkhus, H.B. and Zimmermann, M. (1986) Activities of

single neurons in midbrain and thalamus of cats during conditioned
nocifensive behavior. Pain 26:313-327.

Krisch, B. (1978) Hypothalamic and extrahypothalamic distribution of

somatostatin-immunoreactive elements in the rat brain. Cell tissue res.
195: 499–513.

Lewis, V.A. and G.F. Gebhart (1977) Morphine-induced and stimulation

produced analgesias at coincident periaqueductal gray loci: evaluation of
analgesic congruence, tolerance and cross tolerance. Exp. Neurol. 57: 934
955.

Liebeskind, J.C. and D.J. Mayer (1971) Somatosensory evoked responses in the
mesencephalic central gray matter of the rat. Brain Res. 27: 133-151.
Liu, R.P.C. (1983) Laminar origins of spinal projection neurons to the
periaqueductal gray of the rat. Brain Res. 264: 118-122.
Magoun, H.W., Atlas, D., Ingersoll, E.H. and S.W. Ranson (1937) Associated facial,
vocal and respiratory components of emotional expression: an
experimental study. J. Neurol. Psychopathol. 17: 241-255.
Mantyh, P.W. (1982) The midbrain periaqueductal gray in the rat, cat and
monkey: a Nissl, Weil and Golgi analysis. J. Comp. Neurol. 204; 349-363.
Mantyh, P.W. (1983a) Connections of the midbrain periaqueductal gray in the
monkey. I. Ascending efferent projections. J. Neurophysiol. 49: 567-581.
Mantyh, P.W. (1983b) Connections of the midbrain periaqueductal gray in the
monkey. II. Descending efferent projections. J. Neurophysiol. 49: 582-594.
Marchand, J.E. and N. Hagino (1983) Afferents to the periaqueductal gray in
the rat. A horseradish peroxidase study. Neyrosci. 9: 95-106.
Mayer, D.J. and D.D. Price (1976) Central nervous system mechanisms of
analgesia. Pain 2: 379-404.
Mehler, W.R. (1969) Some neurological species differences -- a posteriori.
Ann. N.Y. Acad. Sci. 167: 424-468.

Meller, S.T. and B.J. Dennis (1986) Afferent projections to the periaqueductal
gray in the rabbit. Neurosci. 19: 927-964.
Millan, M.H., Millan, M.J. and A. Herz. (1986) Depletion of central B-end orphin
blocks midbrain stimulation produced analgesia in the rat. Neurosci. 18:
641-649.

McDougall, A., R.A.L. Dampney and R. Bandler (1985) Cardiovascular

components of the defence reaction evoked by excitation of neuronal cell
bodies in the midbrain periaqueductal gray of the cat. Neurosci. Lett. 60:
69-75.
 2 3

Modianos, D.T. and D.W. Pfaff (1979) Medullary reticular formation lesions and
lordosis behavior in female rats. Brain Res. 171: 334-338.

Morgan, M.M., Depaulis, A. and J.C. Liebeskind (1987) Diazepam dissociates the
analgesic and aversive effects of periaqueductal gray stimulation in the
rat. Brain Res. 423: 395-398.

Moss, M.S. and A.I. Basbaum (1983) The peptidergic organization of the cat
periaqueductal gray. II. The distribution of immunoreactive substance P
and vasoactive intestinal polypeptide. J. Neurosci. 7: 1437-1449.
Nashold, B.S., Wilson, W.P. and D.G. Slaughter (1969) Sensations evoked by
stimulation of the midbrain of man. J. Neurosurg. 30: 14-24.
Nicolaou, N.H., García-Muñoz, M., Arbuthnott, G.W. and D. Eccleston (1979)
Interactions between serotonergic and dopaminergic systems in rat brain
demonstrated by small unilateral lesions of the raphé nuclei. Euro. J.
Pharmacol. 57: 295-305.

Nicoll, R.A., Alger, B.E. and C.E. Jahr (1980) Enkephalin blocks inhibitory
pathways in the vertebrate CNS. Nature 287: 22-25.
Olds, M.E. and J. Olds (1962) Approach escape interactions in rat brain. Ann. J.
Physiol. 203: 803-810.
Oliveras, J.L., Besson, J.M. and G. Guilbaud (1974) Behavioral and
electrophysiological evidence of pain inhibition from midbrain
stimulation in the cat. Exp. Brain Res. 20: 32-44.
Olszewski, J. and D. Baxter (1954) Cytoarchitecture of the Human Brain Stem,
Lippencott, Philadelphia, pp. 50-57.
Peterson, B.W., Pitts, N.G. and K. Fukushima (1979) Reticulospinal connections
with limb and axial motoneurons. Exp. Brain Res. 23: 333-351.
Prieto, G.J., Cannon, J.T. and J.C. Liebeskind (1983) N. raphe magnus lesions
disrupt stimulation-produced analgesia from ventral but not dorsal
midbrain areas in the rat. Brain Res. 261: 53-57.

Romagnano, M.A. and S.A. Joseph (1983) Immunocytochemical localization of

ACTH 1-39 in the brainstem of the rat. Brain Res. 276: 1-16.
Ruda, M.A. (1975) An autoradiographic study of the efferent connections of the
midbrain central gray in the cat. Anat. Rec. 181: 468.
Sakuma, Y. and D. Pfaff (1979a) Facilitation of female reproductive behavior
from mesencephalic central gray in the rat. Am. J. Physiol. 237: R278
R284.

Sakuma, Y. and D. Pfaff (1979b) Mesencephalic mechanisms for integration of

female reproductive behavior in the rat. Am. J. Physiol. 237: R285-R290.
 24

Sakuma, Y. and D. Pfaff (1980a) Convergent effects of lordosis-relevant

somatosensory and hypothalamic influences on central gray cells in the
rat mesencephalon.
Sakuma, Y. and D. Pfaff (1980b) Cells of origin of medullary projections in
central gray of rat mesencephalon. J. Neurophysiol. 44: 1002-1011.
Sanders, K.H., Klein, C.E., Mayer, T.E., Heym, C.H. and H.O. Handwerker (1980)
Differential effects of noxious and non-noxious input on neurons
according to location in ventral periaqueductal grey or dorsal raphe
nucleus. Brain Res. 186: 83-97.

Schenberg, L.C. and F.G. Graeff (1978) Role of the periaqueductal gray
substance in the antianxiety action of benzodiazopines. Pharmacol.
Biochem. Behav. 9: 287-295.

Schmitt, P., Eclancher, F. and P. Karil (1974) Topographic analysis of areas of

negative and positive reinforcement in the central gray matter of the rat.
Physiol. Behav. 12: 271-279.
Schurr, A., Rigor, B.M., Ho, B.T. and N. Dafny (1981) Periaqueductal gray
neurons response to microintophoretically injected morphine in naive
and morphine-dependent rats. Brain Res. Bull. 6: 473–478.
Sharpe, L.G., Garnett, J.E. and T.J. Cicero (1974) Analgesia and hyperreactivity
produced by intracranial microinjections of morphine into the
periaqueductal gray matter of the rat. Behav. Biol. 11: 303-313.
Shen, Z., Yanase, M., Kanosue, K. and T. Nakayama (1986) Effect of
periaqueductal gray morphine injection on thermal response in rats. Jap.
J. Physiol. 36: 485-496.
Shipley, M.T., McLean, J.H. and M.M. Behbehani (1987) Heterogeneous
distribution of neurotensin-like immunore active neurons and fibers in
the midbrain periaqueductal gray of the rat. J. Neurosci. 7: 2025-2034.
Skultety, F.M. (1959) Relation of periaqueductal grey matter to stomach and
bladder motility. Neurology 9: 190-198.
Skultety, F.M. (1962) Circus movements in cats following midbrain stimulation
through chronically implanted electrodes. J. Neurophysiol. 25: 152-164.
Skultety, F.M. (1963) Stimulation of periaqueductal gray and hypothalamus.
Arch. Neurol. 14: 670-680.

Skultety, F.M. and T.M. Gary (1962) Experimental hyperphagia in cats

following destructive midbrain lesions. Neurol. 12: 394-401.
Soper, W.Y. and R. Melzack (1982) Stimulation-produced analgesia: evidence
for somatotopic organization in the midbrain. Brain Res. 251: 301-311.
Spangler, K.M. and B.J. Morley (1987) Somatostatin-like immunoreactivity in
the midbrain of the cat. J. Comp. Neurol. 260: 87-97.
 25

Swett, J.E., McMahon, S.B. and P.D. Wall (1985) Long ascending projections to
the midbrain from cells of lamina I and nucleus of the dorsolateral
funiculus of the rat spinal cord. J. Comp. Neurol. 238: 401-416.
Triepel, J. and C.J.P. Grimmelikhuijzen (1984) A critical examination of the
occurrence of FMRFamide immunoreactivity in the brain of guinea pig
and rat. Histochem. 80: 63-71.

Tseng, L.F., Wei, E.T., Loh, H.H. and C.E. Loh (1980) 3-endorphin: central sites of
analgesia, catalepsy and body temperature changes in rats. J. Pharmacol.
Exp. Ther. 214: 328-332.
Valenstein, E.S. (1965) Independence of approach and escape reactions to
electrical stimulation of the brain. J. Comp. Physiol. Psychol. 60: 20-30.
Wada, J.A., Matsuda, M., Jung, E. and A.E. Hamm (1970) Mesencephalically
induced escape behavior and avoidance performance. Exp. Neurol. 29: 215
220.

Waldbillig, R. (1975) Attack, eating, drinking and gnawing elicited by

electrical stimulation of rat mesencephalon and pons. J. Comp. Physiol.
Psychol. 89: 200-212.
Watanabe, T., Taguchi, Y., Shiosaka, S., Tanaka, J., Kubota, H., Termano, Y.,
Tohyama, M. and H. Wada. (1984) Distribution of the histaminergic neuron
system in the central nervous system of rats: a fluorescent
immunohistochemical analysis with histidine decarboxylase as a marker.
Brain Res. 295: 13-25.

Wiberg, M., Westman, J. and A. Blomqvist (1987) Somatosensory projection to

the mesencephalon: and anatomical study in the monkey. J. Comp. Neurol.
264: 92-117.

Wolfle, T.L., Mayer, D.J., Carder, B. and J.C. Liebeskind (1971) Motivational
effects of electrical stimulation in dorsal tegmentum of the rat. Physiol.
Behav. 7: 569-574.

Yajima, Y., Hayashi, Y. and N. Yoshii (1980) The midbrain central gray
substance as a highly selective neural structure for the production of
ultrasonic vocalization in the rat. Brain Res. 198: 446-452.

Yaksh, T.L., Yeung, J.C. and T.A. Rudy (1976) Systematic examination in the rat
of brainstem sites sensitive to the direct application of morphine:
observation of differential effects within the periaqueductal gray. Brain
Res. 114: 83-103.

Yeung, J.C., Yaksh, T.L. and T.A. Rudy (1977) Concurrent mapping of brain sites
for sensitivity of the direct application of morphine and focal electrical
stimulation in the production of antinociception in the rat. Pain 4: 23-40.
 2 6

CHAPTER II

B-endorphin, enkephalin, dynorphin, and GABA immunoreactivity in

the rat midbrain periaqueductal gray.

David B. Reichling, Yoshiko Kawashima, and Allan I. Basbaum

 27

SUMMARY

It has been suggested that the antinociceptive action of opiates and endogenous
opioid peptides, is mediated, in part, through inhibition of GABAergic neurons in the
midbrain periaqueductal gray matter (PAG). To begin the anatomical study of the
underlying opioid-GABAergic circuitry, we have mapped the distribution of
immunoreactive [-endorphin, enkephalin, dynorphin, and GABA cell bodies and terminals
in the PAG of the rat. Immunoreactivity for each compound forms a distinct pattern of
staining which changes significantly over the rostral-caudal extent of the PAG. In general,
terminal staining for all four putative transmitters is densest near the aqueduct; opioid
terminal staining is also dense in the ventrolateral PAG. Enkephalin and dynorphin
immunoreactive cell bodies cluster in distinct parts of the PAG; GABA-immunoreactive cell
bodies are more widespread. These results suggest that opioid-GABAergic interactions
that contribute to antinociceptive controls are most likely to occur in the medial and
ventrolateral PAG.

INTRODUCTION

Systemically administered opiates initiate powerful descending antinociceptive

controls, in part, through the activation of neurons in the midbrain periaqueductal gray
(PAG) that project to the medullary nucleus raphe magnus (NRM). The PAG
microcircuitry through which this control is generated, however, is poorly understood.
Since the effect of opiates on CNS neurons is exclusively inhibitory, it has been proposed
that opiates do not act directly on the PAG output neuron. Rather, opiates are hypothesized
to disinhibit PAG output neurons, by inhibiting a tonically active inhibitory interneuron
(Yaksh et al. 1976). It has been suggested that the tonically active interneuron is
GABAergic (Basbaum and Fields 1984). This hypothesis is supported by the observations
that PAG microinjection of the GABA antagonist, bicuculline, produces analgesia (Moreau
and Fields 1986) and that microinjection of the GABA agonist, muscimol, attenuates opiate
 28

analgesia (Zambotti et al. 1982). Furthermore, Depaulis et al. (1987) showed that
microinjection of the GABA agonist, THIP, into the ventral PAG, reverses the analgesic
effects of morphine microinjected into the same site. Despite these pharmacological data,
there is no anatomical evidence in support of the proposed endorphin-GABAergic
interactions in the PAG.

A variety of techniques have been used to examine the distribution of GABA

containing cells and terminals in the PAG. One study reported that glutamic acid
decarboxylase- (GAD) immunoreactive neurons are distributed throughout the PAG
(Mugnaini and Oertel 1985); others found that [3H]GABA-accumulating neurons and
neurons immunoreactive for the GABA degradative enzyme, GABA transaminase (GABA
T), are common only in the region of the dorsal raphe (Belin et al. 1979; Nagai et al.
1983). Because of this differing results, and due the fact that the regional distribution of
the different opioid peptide subtypes has not been studied in detail throughout the rostral
caudal extent of the PAG, it is difficult to evaluate the likely location of opioidergic
GABAergic interactions in the PAG.
With the introduction of antisera directed against glutaraldehyde conjugates of the
GABA molecule, it is now possible to more directly examine the distribution of
presumptive GABA-containing cells and terminals. In this study we used a GABA
directed antiserum to study the distribution GABA-immunoreactive neurons and terminals
over the rostral-caudal extent of the PAG. Since our interest is in the relationship of
endorphin elements and GABA neurons, we have also produced detailed maps of
enkephalin, dynorphin and 3-endorphin immunoreactivity in the PAG.
The PAG is spatially heterogeneous with respect to its antinociceptive functions.
For example, stimulation-produced analgesia (SPA) that is elicited from ventral PAG is
reduced by either systemic injection of the opiate antagonist, naloxone, or by lesions of the
nucleus raphe magnus (Cannon et al. 1982; Prieto et al. 1983); SPA from dorsal PAG is
not affected by those manipulations. Furthermore, Fardin et al. (1985) reported that
 29

although electrical stimulation throughout the PAG may produce analgesia, only that
elicited from the DRN and a subregion of the caudal, ventrolateral PAG is free of motor or
aversive side effects. The ventrolateral and medial PAG is also the region from which
microinjection of low doses of opiates elicits the most potent analgeisa (Lewis and Gebhart
1977; Sharpe et al. 1974; Yaksh et al. 1976). To be useful for hypotheses of neural
circuitry, light microscopic cytochemical maps must resolve details at least as small as any
relevant functional zones. For these reasons, the present study describes in detail the
immunocytochemistry of GABA, B-endorphin, enkephalin, and dynorphin at four rostral
caudal levels of the rat midbrain PAG.

METHODS

Male Sprague-Dawley rats (250-300 gram) were used in this study. Three rats
received stereotaxic microinjections of colchicine (100 ug; 6 ul) into the third ventricle in
order to increase staining of cell bodies. Forty-eight hours after administration of
colchicine, the animals were perfused intracardially with 0.1 M phosphate-buffered saline
(PBS) followed by one of two fixative solutions. For opioid peptide
immunocytochemistry the fixative was 4% paraformaldehyde in 0.1 M phosphate buffer
with 4% sucrose, and for GABA immunocytochemistry it was 4% paraformaldehyde and
0.5% glutaraldehyde in 0.05 M PBS. Each rat was perfused with 500 ml of fixative, and
the brain was removed and placed in the same fixative for an additional two hours at 4°C.
Three rats were pretreated with drugs that inhibit GABA uptake; four hours prior to
perfusion, these rats received a third ventricle injection (300 pig in 3 pul) of either 3-alanine
or nipecotic acid. The former drug inhibits neuronal and glial uptake of GABA; the latter
preferentially inhibits neuronal GABA uptake (Johnston 1975; Schon and Kelly 1974).
Thirty- or fifty-micron coronal sections of the rostral pons and midbrain were cut
serially on a freezing microtome or Vibratome. Adjacent sections were immunostained
using the peroxidase-antiperoxidase method (Sternberger 1979). To identify
 30

immunoreactive products of the proenkephalin and prodynorphin precursors we used

antisera directed against leucine enkephalin and dynorphin B, respectively. A few sections
were stained using an antiserum directed against methionine enkephalin-arg-gly-leu; no
significant differences were observed between the distribution of staining for this peptide
and for leucine enkephalin. These antisera were kindly provided by Dr. Ekhard Weber of
the University of Oregon. Radioimmunoassay characterization of each antiserum
demonsumed high specificity for the peptide to which it was directed and almost no cross
reactivity with other peptides (Weber et al. 1982). To identify immunoreactive 3
endorphin we used an antiserum kindly provided by Dr. Volker Höllt of the Max Planck
Institute, FRG. That this antiserum does not cross-react with either dynorphin B or leu
enkephalin is shown by the absence of staining of cell bodies in the PAG of colchicine
treated animals. Furthermore, the pattern of 3-endorphin immunoreactivity was very
different from that seen with the other antisera. Finally, absorption controls were
performed for all three opioid peptide antisera. Adjacent sections were stained with antisera
that had been preabsorbed with the antigen against which the antibody was directed, or
with the one of other opioid peptides. Immunohistochemical staining was only abolished
when the antisera were preabsorbed with 10 puM of the appropriate opioid peptide.
GABA immunoreactivity was mapped using an antiserum kindly provided by Dr.
Andrew Towle of the Cornell University. The antiserum was raised against a
glutaraldehyde-linked GABA-bovine serum albumin (BSA) conjugate; radioimmunoassay
shows no significant cross-reactivity with conjugated glutamate, 3-alanine, or taurine
(Lauder et al. 1986). The antiserum was used at a dilution of 1:3750. Immunoreactivity
in tissue sections was completely abolished by preabsorption of the antiserum with 10 puM
of the GABA-BSA conjugate.
The Paxinos and Watson (1986) atlas of the rat brain was used to deteremine
stereotaxic coordinates and as a reference for mapping the distributions of cell bodies ands
terminals in the PAG. To estimate the proportion of neurons that were GABA
 3 1

immunoreactive, the total number of neurons in the PAG was counted in Nissl-stained
sections adjacent to immunoreacted Sections.

RESULTS

Figures 1-5 show photomicrographs of opioid peptide and GABA immuno

reactivity in sections of the PAG. Figures 6 and 7 schematically illustrate the distributions
of immunoreactive fibers and cell bodies of the four compounds at three rostrocaudal levels
of the PAG. In the following discussion the term, "rostral PAG," refers to levels including
the third cranial nerve (oculomotor) nucleus, "middle" levels of PAG extend from the
caudal pole of the third cranial nerve nucleus to the caudal pole of the superior colliculus,
and "caudal" levels are caudal to the decussation of the superior cerebellar peduncle.

B-Endorphin
There are no [-endorphin-immunoreactive cell bodies in the PAG. In contrast to
the enkephalin and dynorphin fiber staining, which is predominantly punctate in
appearance, 3-endorphin immunoreactivity consists of long, varicose fibers. At caudal
levels of the PAG, 3-endorphin fiber staining extends ventrolaterally, in a band from the
aqueduct beyond the border of the PAG and into the nucleus cuneiformis (figure 1C). A
second zone of 3-endorphin staining, found at all levels of the PAG, extends laterally
outward from the aqueduct, almost to the edge of the PAG. This lateral area of staining
occurs at all levels of the PAG. Only sparse 3-endorphin fibers are found in the dorsal
PAG, ventral PAG, and dorsal raphe nucleus (DRN). The dorsolateral PAG is nearly
devoid of 3-endorphin immunoreactivity.

Enkephalin
Enkephalin-immunoreactive cells are found at all levels of the PAG, but are most
concentrated in the caudal, ventrolateral PAG (figure 2). This cluster of enkephalin cells
 32

does not extend beyond the lateral edge of the PAG. Rostrally, there is another small
cluster of enkephalin-immunoreactive cells dorsal to the aqueduct (figure 3). Moderate
numbers of enkephalin-immunoreactive cells are found in the DRN. The enkephalin
immunoreactive cell bodies are of small-diameter (approximately 12-25 pum). The
population of cells may be classified into the following morphological groups: 70%
fusiform/bipolar, 20% triangular, 10% multipolar. This distribution of cell types is typical
of the general population of neurons in the medial and ventrolateral PAG (Beitz and
Shepard 1985; Liu and Hamilton 1980).
Of the three opioid peptides examined, enkephalin-immunoreactive terminals are the
most dense and most widespread in the PAG. The most intense enkephalin
immunoreactivity generally follows the pattern described for 3-endorphin. In addition,
there are several intense patches of terminal staining unique to enkephalin. One patch is
located in the dorsolateral PAG; it increases in size rostrally (figure 1E). This zone of
dense staining is surrounded by areas of less intense enkephalin immunoreactivity that is
characteristic of most of the dorsolateral PAG. A second bilateral patch of staining occurs
in the lateral part of the DRN (figure 1F). We did not determine if this patch of terminals
overlaps the location of serotonergic DRN neurons. A third notable zone of enkephalin
terminal immunoreactivity is found on the midline, in the ventral part of the DRN; it is
particularly conspicuous rostrally in the caudal linear raphe nucleus (between the
oculomotor nuclei and ventral to Edinger-Westphal (figure 1D). Finally, there is a thin,
subependymal layer of extremely intense enkephalin staining that completely surrounds the
aqueduct.

Dynorphin
Dynorphin-immunoreactive cell bodies are neither as widespread nor as numerous
as enkephalin cells in the PAG. A prominent group of dynorphin-immunoreactive cells
overlaps somewhat with the ventrolateral group of enkephalin cells, but for the most part
 33

they are clustered more ventrally (figure 2). In contrast to enkephalin cells, the cluster of
dynorphin-immunoreactive cells in the ventrolateral PAG is continuous with a group of
dynorphin cells in the nucleus cuneiformis. Another difference between enkephalin and
dynorphin cell staining is the scarcity of dynorphin-immunoreactive cells in the dorsal PAG
(figure 3) and in the DRN (figure 2). Dynorphin-immunoreactive neurons are about the
same size as enkephalin neurons (12-25pum), but they tend to have more primary dendrites:
35% are fusiform/bipolar, 25% triangular, and 40% multipolar. This distribution of cell
types is skewed toward multipolar morphologies compared to the general population of
neurons in the same region of the PAG (Beitz and Shepard 1985; Liu and Hamilton 1980).
The PAG contains moderate levels of dynorphin terminal immunoreactivity. At
caudal and middle levels of the PAG, the regions of most intense dynorphin staining are
similar to the pattern of 3-endorphin staining (figure 1H and I). Rostral to the fourth
cranial nerve nucleus, however, dynorphin immunoreactivity is concentrated dorsolateral
to the aqueduct, where it overlaps a region of dense enkephalin staining (figure 1G). Also
at this level, dynorphin terminal staining is relatively sparse ventrolateral to aqueduct,
where both 3-endorphin and enkephalin staining are the most dense (figure 1G).

GABA

Figure 4 illustrates the distribution of GABA immunoreactivity in the PAG. Figure

4C shows that most GABA-immunoreactive cell bodies are small (1-15 plm diameter) and
exhibit very little dendritic staining. At caudal and middle levels of the PAG, GABA
immunoreactive cell bodies are most common in the ventrolateral and dorsolateral regions,
accounting for approximately 15% of the total Nissl-stained population of neurons; more
rostrally, only the dorsolateral population of intensely stained cells persists in large
numbers. Moderate numbers of GABA-immunoreactive cell bodies are found in the DRN,
particularly in its caudoventral aspect.
 34

Figure 4A shows that GABA terminal staining is found throughout the PAG.
Unlike the varicose appearance of opioid-immunoreactive fibers, GABA-immunoreactive
fibers are punctate (figure 4B). GABA-immunoreactivity is most intense in a thin annulus
that surrounds the aqueduct; this overlaps the region where opioid immunoreactivity is
most dense. Surrounding this annulus, and extending into the ventrolateral PAG, is a
wider zone of somewhat less intense GABA immunoreactivity.
Pretreatment with aminooxyacetic acid, an inhibitor of the GABA-degradative
enzyme, GABA transaminase (GABA-T), produced more intense staining of cell bodies
and proximal dendrites compared to untreated animals. Although the staining in other
regions of the PAG was not changed, the number of labeled cell bodies in the DRN was
increased after GABA-T inhibition (Figure 5). Neither the intensity nor the distribution of
GABA immunoreactivity was changed by pretreatment with nipecotic acid, an inhibitor of
neuronal uptake of GABA (Johnston 1975), or with 3-alanine, an inhibitor of glial uptake
of GABA (Schon and Kelly 1974). Although pretreatment with colchicine might cause
glutamic acid decarboxylase, the GABA synthetic enzyme, to accumulate in cell bodies,
this treatment did not improve staining of cell bodies.

DISCUSSION

The overall pattern of opioid immunoreactivity is consistent with previous

immunocytochemical studies that have mapped 3-endorphin (Finley et al. 1981a),
enkephalin (Fallon and Leslie 1986; Finley et al. 1981b; Sar et al. 1978; Simantov et al.
1977; Uhl et al. 1979), and dynorphin immunoreactivity (Fallon and Leslie 1986;
Khachaturian et al. 1982; Vincent et al. 1982). These studies surveyed large regions of
the brain and thus revealed few of the details within the PAG. The present observations of
enkephalin immunoreactivity in the rat PAG are very similar to our previous description in
the cat (Moss et al. 1983), but disagree with the report by Goia et al. (1988) that
enkephalin staining in rat does not vary along the midbrain axis.
 35

We found that GABA-immunoreactive neurons are very common and widely

distributed in the PAG. Mugnaini and Oertel (1985) reached the same conclusion using an
antiserum directed against GAD. Our results, however, disagree with the reports of
Clemens et al. (1985) and Ottersen and Storm-Mathisen (1984) who used anti-GABA
antisera similar to the one used in this study. The former detected only a dorsal group of
GABA-immunoreactive cells; while the latter reported "scarce" GABA-immunoreacive
neurons in the PAG. It might be relevant that we found GABA-immunoreactive staining
of cell bodies in the PAG to be less intense compared to other regions of the midbrain,
especially the superior colliculus. Since the staining we observed was abolished by
preincubation with a glutaraldehyde conjugate of GABA, it appears to represent genuine
GABA immunoreactivity. Other studies might have understained neurons in the PAG if
other regions of the midbrain were used as the basis for which to control the reaction time
in DAB.

Our results also differ from studies that used other methods to stain putative
GABAergic neurons. Cells that accumulate [3H]GABA or that are stained histochemically
for GABA-T, are found only in the ventromedial PAG and in the DRN (Belin et al. 1979;
Nagai et al. 1983). Interestingly, this is the same region where we observed enhanced
GABA immunoreactivity after pretreatment with AOAA, and AOAA has been used to
increase the accumulation of [3H]GABA by neurons (Schon and Iversen 1972). One
possibility is that this ventral group of putative GABAergic neurons accumulate GABA,
but do not synthesize it. This possibility has also been recognized in the goldfish retina and
in the monkey striate cortex (Kisvárday et al. 1986; Zucker et al. 1984).
The topography of opioid peptide immunoreactivity reflects previously identified
cyotarchitectural subdivisions in the PAG. Hamilton (1973) described a "nucleus medialis"
in the cat, a cell-poor region that surrounds the aqueduct and extends to the ventrolateral
border of the PAG. This cytoarchitectural subdivision corresponds very closely to the
prominent zone of opioid peptide-immunoreactive terminals in the caudal PAG which rings
 36

the aqueduct and has bilateral extensions into the ventrolateral PAG (figure 1C and figure
6C). Beitz (1985) detected a smaller medial subdivision in the rat that contains small cell
bodies that are sparsely distributed. The latter region appears to be coextensive with the
narrow annulus of the most intense opioid peptide immunoreactivity which surrounds the
aqueduct from caudal to mid-levels of the PAG (see figures 1 and 6).
Although the patterns of terminal immunoreactivity are similar for each of the three
opioid peptides studied, there are important differences. 3-endorphin terminals are much
less densely distributed and are found in a far more restricted distribution than the other
opioid peptides. Conversely, enkephalin is easily the densest and most widely distributed
of the opioid peptides examined. Although enkephalin and dynorphin-immunoreactive cell
bodies are located near one another in the lateral PAG, the enkephalin cell bodies are
clustered more dorsally (figure 2). Thus, since dynorphin and enkephalin cell bodies are
spatially segregated, and since 3-endorphin cell bodies are found only in the hypothalamus,
it is unlikely that opioid peptide transmitters coexist in PAG neurons. It follows that axon
terminals containing the three opioid peptides may not only arise from different sources,
but may also have different postsynaptic targets in the PAG.
There has been considerable discussion in the literature over the "mismatch"

between opiate binding sites, which are densest in the dorosolateral part of the rat PAG and
the relatively low level of opioid peptides there (cf. Herkenham and Pert 1982). In fact,
we found a distinct patch of intense enkephalin terminal immunoreactivity in dorsolateral
PAG. It seems somewhat paradoxical, however, that binding is low in the ventrolateral
PAG, where opioid immunoreactivity is dense, and where analgesia is most readily
produced by opiate microinjection (Yaksh et al. 1976). It is also difficult to reconcile the
patterns of opioid peptide staining we observed with the patterns of opioid peptide binding
sites reported by Mansour et al. (1987) using selective pu, 6, and k ligands. For example,
binding of DAGO, a pi-selective ligand, is densest in the dorsolateral PAG, but immuno
reactive 3-endorphin (a putative endogenous pi ligand) is almost absent in the same region.
 37

The selective 6 ligand, DPDPE, does not bind in the PAG, and yet enkephalin
immunoreactive terminals are the most abundant. These data suggest that enkephalin
effects in the rat PAG are not mediated via 8 receptors. Consistent with this idea, Fang et
al. (1986) concluded that an action at 1 receptors is sufficient to explain analgesia produced
by i.c. v. microinjection of putative 6 ligands. This conclusion was based on the similarity
of apparent pA2 values for naloxone antagonism of morphine- and enkephalin analog
induced analgesia. Finally, although both k binding (demonstrated with
[3H]bremazocine), and immunoreactivity for the putative k ligand, dynorphin, are dense
and widely distributed in the PAG, whether k ligands produce analgesia is unclear. In fact,
rather than producing analgesia, i.c.v. dynorphin antagonizes the analgesic action of
morphine or D-ala-enkephalin (Tulunay et al. 1981).
The analgesic action of electrical stimulation in the PAG clearly could result from
direct activation of neurons projecting to the NRM, bypassing any opioid peptidergic
circuitry in the PAG. On the other hand, a component of the analgesia produced by
electrical stimulation or glutamate excitation of cell bodies in the PAG is naloxone
reversible (Akil et al. 1976; Cannon et al. 1982; Oliveras et al. 1977; Urca et al. 1980).
This opioid-mediated component of electrical stimulation might be due to the direct
activation of opioidergic elements in the PAG. Consistent with this idea, potent, aversion
free analagesia can be elicited by electrical stimulation in the ventrolateral PAG (Fardin et
al. 1985), where opioid immunoreactivity is intense. In fact, electrical stimulation of the
PAG produces a naloxone-reversible analgesia and concommitantly depletes pools of 3
endorphin in that region (Millan et al. 1987). Interestingly, levels of enkephalin and
dynorphin in the PAG were unchanged.
If stimulation-produced analgesia is, in fact, due to the direct excitation of
opioidergic elements in the PAG then what is the downstream circuitry that results in
activation of the NRM? Since lesions of the PAG do not produce analgesia (Dostrovsky
and Deakin 1977; Kelly and Glusman 1968; Melzack et al. 1958; Rhodes 1979; Yeung et
 38

al. 1975), it has been proposed that neurons of the PAG that project to the NRM must be
excited in order to produce analgesia. Furthermore, since the only known postsynaptic
effect of opioid peptides is hyperpolarization (Nicollet al. 1980), it has been hypothesized
that opioid peptides do not act directly upon the output (i.e. NRM-projecting) neurons of
the PAG (Yaksh et al. 1976; Basbaum and Fields 1984). Rather, the opioid peptides are
presumed to initiate antinociceptive controls by inhibiting interneurons in the PAG which
tonically inhibit the PAG neurons that project to the NRM. In the hippocampus, opiates
excite pyramidal neurons indirectly, by inhibiting a GABAergic interneuron (Nicollet al.
1980). A comparable opioidergic inhibiton of GABAergic interneurons might occur in the
PAG.

In fact, there is considerable behavioral and pharmacological evidence for analgesia

relevant interactions between GABA and the opioid peptides in the PAG. Nisticó et al.
(1979) reported that microinjection of 3-endorphin into the third ventricle reduced levels of
GABA in the chick brainstem by more than 50%. The hypothesized circuit predicts that
GABA antagonists would mimic the effects of opioids in the PAG, by decreasing
GABAergic transmission. Consistent with this hypothesis, injection of either morphine or
of the GABA antagonists, bicuculline or picritoxin, similarly affect the noxious-stimulus
evoked acitivity of physiologically defined classes of neurons in the NRM (Moreau and
Fields 1986). In addition, intracisternal or PAG microinjection of bicuculline (Ueda et al.
1987; Moreau and Fields 1986) produces potent analgesia.
Conversely, a GABA agonist would be expectd to antagonize opiate-induced
analgesia. In fact, intracerebroventricular, PAG, or DRN microinjection of the GABAA
agonist, muscimol, antagonizes the analgesia produced by systemic morphine or 3
endorphin (Zonta et al. 1981; Zambotti et al. 1982; Romandini et al. 1984). Similarly,
i.c.v. administration of GABA antagonizes the analgesia produced by i.c. v. D-ala”-met
enkephalinamide (Izumi et al. 1980) and inhibition of the uptake or breakdown of GABA
antagonizes systemic morphine analgesia (Ho et al. 1976). Taken together, these results
 39

suggest that analgesia-related actions of GABAergic interneurons in the PAG are mediated
through postsynaptic GABAA receptors.
Since GABA immunoreactivity is intense in regions which contain cells
retrogradely labeled from the NRM, specifically, the lateral and ventrolateral PAG (Abols
and Basbaum 1981; Beitz et al. 1983, Carlton et al. 1983; Fardin et al. 1984; Gallager
and Pert 1978; Reichling et al. 1988), our findings are consistent with the hypothesis that
GABAergic neurons contact the output neurons of the PAG. In fact, we have recently
demonstrated that GABA-immunoreactive terminals make synaptic contact with PAG
neurons that project to the NRM (Reichling and Basbaum 1987, Reichling et al. 1988).
These data support the suggestion that local GABAergic interneurons inhibit the PAG
NRM projection neuron. The other main feature of the proposed circuit, i.e., that opioid
peptide terminals are presynaptic to the GABAergic interneurons remains to be
demonstrated anatomically. To this end, it will be particularly important to determine
whether terminals containing different classes of opioid peptides have different anatomical
relationships with GABA-containing neurons of the PAG. Pharmacological data discussed
above suggest that the three classes of opioid peptides have different effects on PAG
antinociceptive circuitry. It is thus very likely that anatomical studies will reveal significant
differences in their synaptic relationships with GABAergic neurons (and possibly with
PAG output cells).

ACKNOWLEDGEMENTS

The authors thank Ms. Katja Kohler for artistic assistance, and Ms.
Simona Ikeda for photographic assistance. This work was supported by NIH
Public Health Service Grants NS 14627 and 21445.
 40

REFERENCES

Abols, I.A. and Basbaum, A.I., Afferent connections of the rostral medulla of the cat: a
neural substrate for midbrain-medullary interactions in the modulation of pain, J.
Comp. Neurol., 201 (1981) 285-297.
Akil, H., Mayer, D.J. and Liebeskind, J.C., Antagonism of stimulation-produced
analgesia by naloxone, a narcotic antagonist, Science, 191 (1976) 961-962.
Basbaum, A.I. and Fields, H.L., Endogenous pain control systems: brainstem spinal
pathways and endorphin circuitry, Ann. Rev. Neurosci., 7 (1984) 309-338.
Beitz, A.J., M.A. Mullet and Weiner, L.L., The periaqueductal gray projections to the rat
spinal trigeminal, raphe magnus, gigantocellular pars alpha and paragigantocellular
nuclei arise from separate neurons, Brain Res. 288 (1983) 307-314.
Beitz, A.J., The midbrain periaqueductal gray in the rat. I. Nuclear volume, cell density,
orientation, and regional subdivisions, J. Comp. Neurol., 237 (1985) 445-459.
Beitz, A.J. and Shepard, R.D., The midbrain periaqueductal gray in the rat. II. A Golgi
analysis, J. Comp. Neurol., 237 (1985) 460-475.
Belin, M.F., Aguera, M., Tapaz, A., McRae-DeQueurce, A., Bobillier, P. and Pujol,
J.F., GABA-accumulating neurons in the nucleus raphe dorsalis and periaqueductal
gray in the rat: a biochemical and radioautographic study, Brain Res., 170 (1979)
279-297.

Cannon, J.T., Prieto, G.J. and Liebeskind, J.C., Evidence for non-opioid forms of
stimulation-produced analgesia in the rat, Brain Res., 243 (1982) 315-321.
Carlton, S.M., G.R. Leichnetz, E.G. Young and D.J. Mayer (1983) Supramedullary
afferents of the nucleus raphe magnus in the rat: a study using the transcannula
HRP gel and autoradiographic techniques. J. Comp. Neurol. 214:43-58.
Clements, J.R., Beitz, A.J., Larson, A.A. and Madl, J.E., The ultrastructural localization
of polyclonal GABA and monoclonal glutamate immunoreactivity in the rat midbrain
periaqueductal gray, Soc. Neurosci. Abs., 11 (1985) 126.
Depaulis, A., Morgan, M.M. and Liebeskind, J.C., GABAergic modulation of the
analgesic effects of morphine microinjected in the periaqueductal gray matter of the
rat, Brain Res., 436 (1987) 223-228.
Dostrovsky, J.O. and Deakin, J.F.W., Periaqueductal grey lesions reduce morphine
analgesia in the rat, Neurosci. Lett., 4 (1977) 99-103.
Fallon, J.H. and Leslie, F.M., Distribution of dynorphin and enkephalin peptides in the rat
brain, J. Comp. Neurol. 249 (1986) 293-336.
Fang, F.G., Fields, H.L. and Lee, N.M., Action at the mu receptor is sufficient to explain
the supraspinal analgesic effect of opiates, J. Pharmacol. Exp. Ther., 238 (1986)
1039-1044.
 41

Fardin, V., Oliveras, J.L. and Besson, J.M., Projections from the periaqueductyal gray
matter to the B3 cellular area (nucleus raphe magnus and nucleus reticularis
paragigantocellularis) as revealed by the retrograde transport of horseradish
peroxidase in the rat, J. Comp. Neurol. 223 (1984) 483-500.
Fardin, V., Oliveras, J.L. and Besson, J.M., A reinvestigation of the analgesic effects
induced by stimulation of the periaqueductal gray matter in the rat. II. Differential
characteristics of the analgesia induced by ventral and dorsal PAG stimulation, Brain
Res., 306 (1985) 125-139.
Finley, J.C.W., Lindström, P. and Petrusz, P., Immunocytochemical localization of B
endorphin-containing neurons in the rat brain, Neuroendocrinol. 33 (1981a) 28-42.
Finley, J.C.W., Maderdrut, J.L. and Petrusz, P., The immunocytochemical localization of
enkephalin in the central nervous system of the rat. J. Comp. Neurol. 198 (1981b)
541–565.

Gallager, D.W. and Pert, A., Afferents to brain stem nuclei (brain stem raphe, nucleus
reticularis pontis caudalis and nucleus gigantocellularis) in the rat as demonstrated
by microiontophoretically applied horseradish peroxidase. Brain Res. 144 (1978)
257-275.

Goia, M. and Bianchi, R., The distribution of substance P and met-enkephalin in the
periaqueductal gray matter of the rat. Bas. Appl. Histochem. 32 (1988) 103-108.
Hamilton, B.L., Cytoarchitectural subdivisions of the periaqueductal gray matter in the cat,
J. Comp. Neurol., 149 (1973) 1-28.
Herkenham, M. and Pert, C.B., Light microscopic localization of brain opiate receptors: a
general autoradiographic method which preserves tissue quality, J. Neurosci., 2
(1982) 1129-1149.
Ho, I.K., Loh, H.H. and Way, E.L., Pharmacological manipulation of gamma
aminobutyric acid (GABA) in morphine analgesia, tolerance and physical
dependence, Life Sci., 18 (1976) 1111-1124.
Izumi, K., Munekata, E., Yamamoto, H., Nakanishi, T. and Barbeau, A., Effects of
taurine and Y-aminobutyric acid on akinesia and analgesia induced by D-Ala”-met
enkephalinamide in rats, Peptides, 1 (1980) 139-146.
Johnston, G.A.R., Krogsgaard-Larsen, P. and Stephanson, A., Betel nut constituents as
inhibitors of Y-aminobutyric acid uptake, Nature, 258 (1975) 627-628.
Kelly, D.D. and Glusman, M., Aversive thresholds following midbrain lesions, J. Comp.
Physiol. Psychol., 66 (1968) 25-34.
Khachaturian, H., Watson, S.J., Lewis, M.E., Coy, D., Goldstein, A. and Akil, H,
Dynorphin immunocytochemistry in the rat central nervous system, Peptides, 3
(1982) 941-954.
 42

Kisvárday, Z.F., Cowey, A., Hodgson, A.J. and Somogyi, P., The relationship between
GABA immunoreactivity and labelling by local uptake of [3H]GABA in the striate
cortex of monkey, Exp. Brain Res.62 (1986) 89-98.
Lauder, J.M., Han, V.K.M., Henderson, P., Verdoon, T. and Towle, A.C., Prenatal
ontogeny of the GABAergic system in the rat brain: an immunocytochemical study,
Neurosci, 19 (1986) 465-493.
Lewis, V.A. and Gebhart, G. F., Morphine-induced and stimulation-produced analgesias
at coincident periaqueductal gray loci: evaluation of analgesic congruence, tolerance,
and cross-tolerance, Expt. Neurol., 57 (1977) 934-955.
Liu, R.P.C. and Hamilton, B.L., Neurons of the periaqueductal gray matter as revealed by
Golgi study, J. Comp. Neurol., 189 (1980) 403-418.
Mansour, A., Khachaturian, H., Lewis, M.E., Akil, H. and Watson, S.J.,
Autoradiographic differentiation of mu, delta, and kappa opioid receptors in the rat
forebrain and midbrain, J. Neurosci., 7 (1987) 2445–2464.
Melzack, R., Stotler, W.A. and Livingston, W.K., Effects of discrete brainstem lesions in
cats on perception of noxious stimulation, J. Neurophysiol., 21 (1958) 353–367.
Millan, M.J., Czlonkowski, A., Millan, M.H. and Herz A., Activation of periaqueductal
grey pools of 3-endorphin by analgetic electrical brain stimulation in freely moving
rats, Brain Res., 407 (1987) 199-203.
Moreau, J.-L. and Fields, H.L., Evidence for GABA involvement in midbrain control of
medullary neurons that modulate nociceptive transmission, Brain Res., 397 (1986)
37–46.

Moss, M.S., Glazer, E.J. and Basbaum, A.I., The peptidergic organization of the cat
periaqueductal gray. I. The distribution of immunoreactive enkephalin-containing
neurons and terminals, J. Neurosci., 3 (1983) 603-616.
Mugnaini, E. and Oertel, W.H., An atlas of the distribution of GABAergic neurons and
terminals in the rat CNS as revealed by GAD immunohistochemistry, In A.
Björklund and T. Hökfelt, (eds.), Handbook of Chemical Neuroanatomy, Vol. 4,
GABA and Neuropeptides in the CNS. Part I, Elsevier, Amsterdam, 1985, pp. 436
608.

Nagai, T., McGeer, T.L., and McGeer, E.G., Distribution of GABA-T intensive neurons
in the rat forebrain and midbrain, J. Comp. Neurol., 218 (1983) 220-238.
Nicoll, R.A., Alger, B.E. and Jahr, C.E., Enkephalin blocks inhibitory pathways in the
vertebrate CNS, Nature, 287 (1980) 22-25.
Nisticó, G., DiGiorgio, R.M., Rotiroti, D., Naccari, F. and Calatroni, A., Effects of
intraventricular B-endorphin on GABA system in some areas of chick brain, Res.
Comm. Chem. Pathol. Pharmacol., 26 (1979) 469-478.
 4 3

Oliveras, J.L., Hosobuchi, Y., Redjemi, F., Guilbaud, G. and Besson, J.M., Opiate
antagonist, naloxone, strongly reduces analgesia induced by stimulation of a raphe
nucleus (centralis inferior), Brain Res., 120 (1977) 221-229.
Ottersen, O.P. and Storm-Mathisen, J., Glutamate- and GABA-containing neurons in the
mouse and rat brain, as demonstrated with a new immunocytochemical technique, J.
Comp. Neurol., 229 (1984) 374-392.
Paxinos, G. and Watson, C., The Rat Brain in Stereotaxic Coordinates, 2nd ed.,
Academic Press, Orlando, Fl: 1986.
Prieto, G.J., Cannon, J.T. and Liebeskind, J.C., N. raphe magnus lesions disrupt
stimulation-produced analgesia from ventral but not dorsal midbrain areas in the rat,
Brain Res., 261 (1983) 53-57.
Reichling, D.B., Cho, H.J. and Basbaum, A.I., GABA-immunoreactive synaptic contacts
onto projection neurons of the periaqueductal gray matter and the nucleus raphe
magnus, Soc. Neurosci. Abs., 14 (1988) 856.
Reichling, D.B. and Basbaum, A.I., Anatomical evidence that GABA influences the
projection from the periaqueductal gray matter to the nucleus raphe magnus, Pain,
Supp. 4 (1987) S40.
Rhodes, D.L., Periventricular system lesions and stimulation-produced analgesia, Pain, 7
(1979) 51-63.
Romandini, S. and Samanin, R., Muscimol injections in the nucleus raphe dorsalis block
the antinociceptive effects of morphine in rats: apparent lack of 5-hydroxytryptamine
involvement in muscimol's effect, Br. J. Pharmacol., 81 (1984) 25-29.
Sar, M., Stumpf, W.E., Miller, R.J., Chang, K.-J. and Cuatrecasas, P., Immunohisto
chemical localization of enkephalin in rat brain and spinal cord, J. Comp. Neurol.
182 (1978) 17-38.
Schon, F. and Kelly, J.S., The characterisation of [3H]GABA uptake into the sattelite glial
cells of rat sensory ganglia, Brain Res., 66 (1974) 289-300.
Sharpe, L.G., Garnett, J.E. and Cicero, T.J., Analgesia and hyperreactivity produced by
intracranial microinjection of morphine into the periaqueductal gray matter of the rat,
Behav. Biol., (1974) 303-313.
Simantov, R., Kuhar, M.J., Uhl, G.R. and Snyder, S.H., Opioid peptide enkephalin:
Immunohistochemical mapping in rat central nervous system, Proc. Natl. Acad. Sci.,
74 (1977) 2167–2171.
Sternberger, L.A., Immunocytochemistry, 2nd ed., John Wiley and Sons, New York,
1979.

Tulunay, F.C., Jen, M.-F., Chang, J.-K., Loh, H.H. and Lee N.M., Possible regulatory
role of dynorphin on morphine- and beta-endorphin-induced analgesia, J. Pharmacol.
Exp. Ther., 219 (1981) 296–298.
 44

Ueda, H., Ge, M., Satoh, M. and Takagi, H., Subconvulsive doses of intracisternal
bicuculine methiodide, a GABAA receptor antagonist, produce potent analgesia as
measured in the tail pinch test in mice, Euro. J. Pharmacol., 136 (1987) 129-131.
Uhl, G.R., Goodman, R.R., Kuhar, M.J. Childers, S.R. and Snyder, S.H.,
Immunohistochemical mapping of enkephalin containing cell bodies, fibers and nerve
terminals in the brain stem of the rat. Brain Res. 166 (1979) 75-94.
Urca, G., Nahin, R.L. and Liebeskind, J.C., Glutamate-induced analgesia: blockade and
potentiation by naloxone. Brain Res. 192 (1980) 523-530.
Vincent, S.R., Hökfelt, T., Christensson, I. and Terenius, L., Dynorphin-immunoreactive
neurons in the central nervous system, Neurosci. Lett., 33 (1982) 185-190.

Weber, E., Evans, C.J., Chang, J.-K. and Barchas, J.D., Antibodies specific for o-N-
acetyl-3-endorphins: radioimmunoassays and detection of acetylated 3-endorphins
in pituitary extracts, J. Neurochem., 38 (1982) 436-447.
Yaksh, T.L., Yeung, J.C. and Rudy T.A., Systematic examination in the rat of brainstem
sites sensitive to the direct application of morphine: observation of differential effects
within the periaqueductal gray, Brain Res., 114 (1976) 83-103.
Yeung, J.C., Yaksh, T.L. and Rudy, T.A., Effects of brain lesions on the antinociceptive
properties of morphine in rats, Clin. Exp. Pharmacol. Physiol., 2 (1975) 261-268.
Zambotti, F., Zonta, N., Parenti, M., Tommasi, R., Vicentini, L., Conci, F. and
Montegazza, P., Periaqueductal gray matter involvement in the muscimol induced
decrease of morphine antinociception, Naunyn-Schmiedeberg's Arch. Pharmacol.,
318 (1982) 368-369.
Zonta, N., Zambotti, F., Vicentini, R., Tammiso, R. and Mantegazza, P., Effects of some
GABA-mimetic drugs on the antinociceptive activity of morphine and fl-endorphin in
rats, Naunyn-Schmiedeberg's Arch. Pharmacol., 316 (1981) 231-234.
Zucker, C., Yazulla, S. and Wu, J.Y., Non-correspondence of [3H]GABA uptake and
GAD localization in goldfish amacrine cells, Brain Res. 298 (1984) 154-158.
 45

Fig. 1. These color darkfield photomicrographs illustrate the distribution of 3-endorphin

(A,B,C), enkephalin- (D,E,F), and dynorphin- (G,H,I) immunoreactive fibers in the PAG.
In this darkfield illumination, immunoreactive fibers appear golden-orange. The top row of
plates shows the rostral PAG at the level of the third nerve nucleus; the middle row shows
the PAG at the level of the fourth nerve nucleus; the bottom row shows the caudal PAG at
the level of the decussation of the superior cerebellar peduncle. Arrowheads indicate
distinctive features of the staining patterns, which are described in the text.
 47

Fig. 2. These low-magnification bright-field photomicrographs illustrate the patterns of

opioid-immunoreactive cell bodies in the caudal PAG. The left-hand plate shows
enkephalin-immunoreactive cell bodies (arrowheads) in the verntrolateral PAG.
Dynorphin-immunoreactive cell bodies, shown in the right-hand plate, are also found in the
ventrolateral PAG (arrowheads), but they are located farther from the aqueduct and more
ventrally than are enkephalin cells. The cluster of dynorphin cells also extends
ventrolaterally into the nucleus cuneiformis (arrow). Scale bar = 0.5 mm.
#O
 49

Fig. 3. These photomicrographs show opioid-immunoreactive cell bodies in the dorsal

PAG, at the level of the fourth nerve nucleus. Enkephalin-immunoreactive neurons (left
hand plate) are found in bilateral clusters dorsal to the aqueduct (AQ), but are rarely found
on the midline. Dynorphin-immunoreactive cell bodies (right-hand plate) are rare in the
dorsal PAG. Scale bar = 100 pum.
 5 1

Fig. 4. These photomicrograhs show GABA-immunoreactive elements in the PAG. A.

Low-magnification photomicrograph of GABA-immunoreactive fibers in the PAG at the
level of the decussation of the superior cerebellar peduncle (Bright-field illumination was
used since dark-field illumination does not satisfactorily represent variations in the intensity
of GABA fiber staining). GABA-immunoreactive fibers are most dense in the ventrolateral
PAG, near the aqueduct (arrow). Scale bar = 1 mm. B. This higher-magnification
photomicrograph of a region ventrolateral to the aqueduct (AQ) shows that individual
GABA-immunoreactive fibers appear as numerous small, dark punctae (arrowheads).
Scale bar = 100 pum. C. This plate shows GABA-immunoreactive cell bodies (arrows) in
the dorsolateral PAG at the level of the fourth nerve nucleus. Scale bar = 100 pum.
 53

Fig. 6. Schematic illustration of the patterns of 3-endorphin- and enkephalin

immunoreactive fibers and cell bodies at three rostrocaudal levels of the PAG. The density
of cross-hatching in the left half of each drawing represents the relative density of
immunoreactive fibers. Black dots in the right half of each figure represent 5
immunoreactive cell bodies in a typical 50pum section. The drawings are adapted from
Paxinos and Watson (1986); numbers indicate the distance rostral to interaural zero
according to that atlas.
 55

Fig. 7. Schematic illustration of the patterns of dynorphin- and GABA-immunoreactive

fibers and cell bodies at three rostrocaudal levels of the PAG. The density of cross
hatching in the left half of each drawing represents the relative density of immunoreactive
fibers. Black dots in the right half of each figure represent 5 immunoreactive cell bodies in
a typical 50 plm section. -
 57

CHAPTER III

Axon Collaterals to the Hypothalamus and Thalamus

from Periaqueductal Gray Neurons that Project to Raphe Magnus

David B. Reichling and Allan I. Basbaum

 58

SUMMARY

Antinociceptive effects elicited from the midbrain may involve ascending, as well as
descending, projections from the periaqueductal gray matter (PAG) and dorsal raphe
nucleus (DRN). To investigate the relationship between these efferent pathways we
performed a double-labeling retrograde tracer study using WGAapoHRP-Au and
fluorescent tracers. One tracer was microinjected in the medullary nucleus raphe magnus
(NRM), while another tracer was injected into one of several regions rostral to the PAG
which have been implicated in nociceptive or antinociceptive processses. The results can
be grouped into two categories according to the patterns of labeled neurons in the PAG and
DRN. First, injections into the ventrobasal thalamus, lateral hypothalamus, amygdala, and
cerebral cortex labeled neurons in the DRN but not in other regions in the PAG. Between
85 and 90% of these projection neurons were 5-HT-immunoreactive, and between 8 and
17% were also retrogradely labeled from the RVM. Second, only injections into the
ventrobasal hypothalamus or medial thalamus labeled neurons in the PAG itself. The latter
neurons were most numerous in the ventrolateral, lateral, and dorsal parts of the ipsilateral
PAG. Injections in the medial thalamus, but not in the ventrobasal hypothalamus, also
labeled neurons in the DRN. Between 13 and 20% of the neurons retrogradely labeled
from these regions were also retrogradely labeled from the RVM. The existence of these
rostrally-directed collaterals of the PAG-NRM projection complicate the interpretation
electrical brain stimulation studies, and suggest the possibilty that the medial thalamus and
ventrobasal hypothalamus may be subject to antinociceptive controls ascending from the
PAG.

INTRODUCTION

Most studies that address the antinociceptive function of the midbrain

periaqueductal gray matter (PAG) emphasize its descending connections with neurons in
the nucleus raphe magnus (NRM) of the rostral ventromedial medulla (RVM). Although
 59

this view of the PAG as the upper tier in a descending antinociceptive system (Fields and
Basbaum 1978; Basbaum and Fields 1984) has proved very useful in the study of
bulbospinal inhibitory pathways, the PAG also has extensive rostral projections. For
example, anterograde tracing studies demonstrated PAG projections to: hypothalamus
(periventricular gray, dorsal area, dorsomedial, posterior, anterior, lateral, medial,
ventromedial, mammillary, and supramammillary nuclei), thalamus (paracentral,
centrolateral, paracentral, parafascicular, reticular, central medial, mediodorsal, rhomboid,
and reuniens nuclei, and the ventrobasal complex), zona incerta, fields of Forel, pretectal
area, and lateral habenula (Chi 1970; Hamilton 1973; Mantyh 1983a).
Since the PAG receives a variety of nociceptive inputs from the spinal cord dorsal
horn and brainstem reticular formation, it has been suggested that ascending efferents from
the PAG primarily relay nociceptive information (cf. Mehler et al. 1969). Consistent with
this idea, electrical stimulation of the PAG, in a number of species, including humans not
only produces antinociception, but also a variety of pain-related reactions including
hyperalgesia, vocalization, defense, fear, and panic, (Fardin et al. 1984b; Schmidek et al.
1971; Nashold et al. 1969). On the other hand, analgesia can be elicited from the PAG and
from many of its more rostral targets either by electrical stimulation or narcotic
microinjection. Conceivably, ascending projections from the PAG may contribute to the
antinociceptive effects elicited by manipulations in the PAG itself. Such ascending controls
might mediate effects such as anti-aversion or disruption of integrated reactions to painful
stimuli.

The present study addresses the organization of ascending projections from the
PAG and to relate these projections to the better characterized "antinociceptive" projection
to the NRM. In fact, some PAG neurons have axons that collaterize to both sides of
the thalamus (Barbaresi et al. 1982), and some neurons in the PAG that project to the
NRM collateralize to other parts of the medulla (Beitz et al. 1983). Specifically, we
used retrograde tracers to determine if the PAG-RVM pathway collateralizes to regions of
 60

the forebrain that have been implicated in antinociceptive processes: the thalamic
parafascicular nucleus, ventroposterior thalamus, ventrobasal hypothalamus, lateral
hypothalamus, amygdaloid complex, medial or lateral frontal cortex, parietal sensorimotor
cortex, or visual occipital cortex. Since the DRN has also been implicated in
antinociception (Fardin et al. 1985b) and also projects to the NRM (in the rat) and to some
of these forebrain areas, we assessed the contribution of dorsal raphe neurons to these
pathways using 5-HT immunocytochemistry. A preliminary report of the results has been
published (Reichling and Basbaum 1986)

METHODS

Adult male Sprague-Dawley rats (Bantin Kingman) were used in this study. Under
pentobarbitol anesthesia the animals were placed in a stereotaxic frame, the skull was
exposed, and a hole was drilled over the appropriate target for the tracer injection. The
Stereotaxic atlas of Paxinos and Watson (1986) was used to determine coordinates.
Fourteen rats first received one stereotaxic microinjection (0.10-0.15 pul) of a retrograde
flourescent tracer, either 5% aqueous True Blue (Sigma) or 5% Fluorogold in sterile saline
(Fluoroprobe; Schmued and Fallon 1986), into the RVM, the injection was centered in the
NRM. A second retrograde tracer, the colloidal gold-protein complex, WGAapoHRP-Au
(Basbaum and Menetrey 1987), was then microinjected (0.05-0.15 pul) into one site rostral
to the PAG. The rostral target was either the thalamic parafascicular nucleus,
ventroposterior thalamus, ventrobasal hypothalamus, lateral hypothalamus, amygdaloid
complex, medial or lateral frontal cortex, parietal (sensorimotor) cortex, or occipital (visual)
COrtex.

To determine if smaller injections restricted to the NRM would produce results

different from the larger injections of flourescent tracer, three rats received the
WGAapoHRP-Au tracer injection into the RVM. In these rats, the fluorescent tracer was
injected into medial thalamus, ventromedial hypothalamus, or cortex.
 61

Three to fourteen days after injection of the tracers, the rats were deeply
anesthetized with pentobarbitol and perfused transcardially with 100 ml 0.1 M phosphate
buffered saline (PBS) followed by 500 ml of 4% paraformaldehyde in PBS. The brains
were removed, cryoprotected in 30% sucrose PBS, and sectioned on a freezing microtome.
The PAG was sectioned at 25 or 30 pum thickness; each injection site was sectioned at 100
|Im.

Alternate sections of the PAG were silver-enhanced to make the colloidal gold tracer
visible at the light microscopic level (Basbaum and Menetrey 1987). These sections were
then mounted on uncoated glass slides and coverslipped with a medium that retards fading
of fluorescent tacers (Dabco, Aldrich). By using a combination of ultraviolet epi
fluorescent illumination, and transmitted dark-field illumination, it was possible to
simultaneously view the fluorescent tracers and the silver-enhanced WGAapoHRP-Au.
In some experiments, silver-enhanced sections of the PAG were stained by
immunofluorescence for 5-hydroxytryptamine (5-HT). These sections were first incubated
for 1/2 hr in a blocking solution of 3% normal goat serum (NGS) and 0.3% triton X-100 in
PBS. The sections were next incubated overnight in rabbit anti-5-HT antiserum (IncStar)
diluted 1:3000 in 196 NGS and 0.3% Triton X-100 in PBS. After several washes in PBS,
the sections were incubated for 1 hour in flourescein conjugated goat anti-rabbit IgG
(Cappel), diluted 1:100. After washing, the sections were mounted and coverslipped. The
locations of single- and double-labeled cells were plotted on camera lucida drawings.

RESULTS

Labeling from rostral ventromedial medulla injections

Tracer injections into the RVM consistently labeled neurons in the ventral,
ventrolateral, lateral, and dorsal PAG. Very few retrogradely labeled neurons were found
in the dorsolateral PAG. At levels of the PAG rostral to the fourth cranial nerve nucleus the

dorsal group of cells persists, while cells in other parts of the PAG diminish in number.
 62

Retrogradely labeled neurons in the dorsal raphe nucleus (DRN) were more common in the
lateral part of the nucleus. The same pattern of retrograde labeling in the PAG was found
when either WGAapoHRP-Au or fluorescent tracers were used. These results agree with
previous reports of retrograde tracing experiments performed in rat (Beitz et al. 1983;
Carlton et al. 1983; Gallager and Pert 1978) cat (Abols and Basbaum 1981) and monkey
(Chung et al. 1983). However, Fardin et al. (1984a) reported that neurons
retrogradely labeled from the NRM were found only in the lateral DRN and dorsal to
the aqueduct in the PAG of the rat. We found that these two regions where the
largest-diameter (approximately 30 pum) retrogradely labeled cell bodies
Labeling from ventrobasal hypothalamus injections
Injection sites in the ventrobasal (VB) hypothalamus (figure 1) included the arcuate
nucleus and ventromedial nuclei; there was usually some spread of tracer in the needle tract
into the dorsomedial nucleus of the hypothalamus. The pattern of labeling produced by
these injections was very similar to the distribution of neurons in the PAG that were
retrogradely labeled from the RVM. Two thirds of the retrogradely labeled neurons were
found ipsilateral to the injection site. Labeled neurons (figure 5) were found in the
ventrolateral, lateral, and dorsal PAG, but few were found in the dorsolateral PAG. The
dorsal group of cells was most prominent at rostral levels of the PAG. Finally, the DRN
contained few labeled neurons -- less than 2% of neurons retrogradely labeled from the VB
hypothalamus were 5-HT-immunoreactive.
Figure 2 illustrates examples of single- and double- labeled neurons.
Approximately 13% of cell bodies retrogradely labeled from the VB hypothalamus were
also labeled from the RVM. These double-labeled neurons appeared to be evenly
distributed among the single-labeled neurons.
Labeling from lateral hypothalamus injections
Tracer injections in the lateral hypothalamus (figure 1) were limited to the lateral
hypothalamic area. Neurons retrogradely labeled from these injections were found only in
 63

the DRN. An example of 5-HT-immunoreactive retrogradely labeled neurons is shown in

figure 4. Most retrogradely labeled neurons were found in the medial DRN, and in the
lateral parts of the nucleus ipsilateral to the injection site. The more laterally situated cells
are most common at mid rostrocaudal levels of the DRN (at the level of the fourth cranial
nerve nucleus); at more rostral levels, retrogradely labeled cells are found only on the
midline; 89% of the retrogradely labeled neurons were also 5-HT-immunoreactive.
Approximately 10% of neurons retrogradely labeled from the lateral hypothalamus were
also retrogradely labeled by tracer injections in the RVM. These double-labeled neurons
were located randomly among the population of DRN neurons retrogradely labeled from
the lateral hypothalamus.
Labeling from medial thalamus injections
Tracer injections in the medial thalamus (figure 1) were centered in the nucleus
parafascicularis (Pf). Although the passage of the fasciculus retroflexus through the medial
part of Pf is a convenient histological landmark, nuclear boundaries in the rat medial
thalamus are somewhat difficult to distinguish. It is, thus, likely that some tracer spread
into adjacent medial thalamic nuclei, particularly the central medial nucleus (CM) and
centrolateral nucleus (CL) of the internal medullary lamina.
Approximately 75% of neurons retrogradely labeled from the medial thalamus were
located in the PAG ipsilateral to the injection site (see figure 3, plate A, and figure 4). All
regions of the PAG, except for the rostral dorsolateral sector, contained large numbers of
labeled neurons. This distribution was similar to that produced by injections in the
ventrobasal hypothalamus. In contrast to ventrobasal hypothalamus injections, however,
wefound many more retrogradely labeled neurons in the DRN -- after medial thalamus
injections approximately 6% of all neurons labeled from the medial thalamus were
immunoreactive for 5-HT

Approximately 20% of cell bodies retrogradely labeled from the medial thalamus
were also labeled by injections into the RVM. Unlike the results after injections into the
 64

ventrobasal hypothalamus, these double-labeled cell bodies are not evenly distributed
among the population of single-labeled neurons. Among neurons retrogradely labeled from
the medial thalamus, 22% were double-labeled on the side ipsilateral to the injection site;
10% were double-labeled on the contralateral side; the latter were most common in the

lateral part of the PAG. There was also a pronounced change in the proportion of double
labeled neurons along the rostrocaudal axis of the PAG. At the most caudal levels of the
PAG only about 4% of cell bodies labeled from the medial thalamus were also labeled from
the RVM; at the level of the fourth cranial nerve nucleus this number was greater than 30%.
Labeling from lateral thalamus injections
Tracer injections in the lateral thalamus (figure 1) were centered in the
ventroposterior nuclei lateralis and medialis, but a small amount of tracer may have spread
medially into the posterior nuclear group, or dorsally along the needle track into the lateral
posterior and laterodorsal nuclei. These injections labeled a small number of neurons
which were located in the DRN (figure 5); 85% of these neurons were 5-HT
immunoreactive. In the caudal DRN retrogradely labeled neurons were most common in
the ventromedial and lateral parts of the nucleus. At the level of the fourth cranial nerve
nucleus, the location of labeled neurons shifted to more dorsal parts of the DRN.
Approximately 17% of neurons labeled from the lateral thalamus were also labeled from the
RVM, and these double-labeled neurons appeared to be distributed evenly among the
population of neurons labeled from the lateral thalamus.
Labeling from telencephalic injections
Tracer injections placed in the medial (figure 1) or lateral parts of the frontal cortex,
amygdala, parietal cortex, and occipital cortex all produced very similar results.
Retrogradely labeled neurons were found only in the DRN; between 87 and 93% were 5
HT-immunoreactive. In the caudal DRN, caudal to the decussation of the superior
cerebellar peduncle, retrogradely labeled neurons were most common in the medial part of
the DRN; at mid rostrocaudal levels of the DRN, labeled neurons were more common in
 65

the lateral parts of the nucleus. The proportion of neurons projecting to these regions that
were double-labeled by injections in the RVM ranged between 8 and 17%. Double-labeled
neurons did not cluster within the distributions of Single-labeled neurons.

DISCUSSION

Previous studies have demonstrated that single PAG neurons can have axon
collaterals in two nearby areas of the brain (Barbaresi et al. 1982; Beitz et al. 1983), and
the present results show that some PAG axons send collaterals in opposite, rostral and
caudal, trajectories. Our results can be grouped into two categories according to the
patterns of labeled neurons in the PAG and DRN after forebrain injections of retrograde
tracer. First, injections into the ventrobasal thalamus, lateral hypothalamus, amygdala, and
cerebral cortex labeled neurons in the DRN but not in other regions in the PAG. Between
85 and 90% of these projection neurons were 5-HT-immunoreactive, and between 8 and
17% of the total were also retrogradely labeled from the RVM. Second, only injections
into the ventrobasal hypothalamus or medial thalamus labeled neurons in the PAG itself.
These labeled neurons were most numerous in the ventrolateral, lateral, and dorsal parts of
the ipsilateral PAG. Injections in the medial thalamus, but not in the VB hypothalamus,
also labeled neurons in the DRN. Between 13 and 20% of the neurons retrogradely labeled
from these regions were also retrogradely labeled from the RVM.
The presence of neurons in the PAG and DRN that send axon collaterals to the
RVM and also to regions rostral to the PAG is a major complication for studies that use
electrical brain stimulation to investigate antinociceptive systems. Electrical stimulation of
areas rostral to the PAG, especially the medial thalamus and ventrobasal hypothalamus,
could antidromically activate axons that originate from cell bodies in the PAG and DRN.
Consequently, after reaching the PAG, action potentials could be propogated
orthodromically in axon branches that terminate in the RVM. Based on these results, it
Seems that the use of electrical stimulation could lead to the incorrect conclusion that under
 66

physiological conditions, the neuronal circuitry of these rostral regions can activate
descending antinociceptive controls. Of course, stimulation-produced analgesia (SPA)
elicited from the RVM might also be complicated by electrical activation of these bifurcating
axons. Specifically, action potentials travelling antidromically from the NRM to the PAG
could activate PAG projections to a variety of rostral targets.

MEDIAL DIENCEPHALON INJECTIONS

The PAG, medial thalamus, and ventromedial hypothalamus are part of a

continuous periventricular zone in which morphine microinjection elicits analgesia (Herz et
al. 1970; Mayer and Price 1976; Pert and Yaksh 1974; Yaksh et al. 1976; Yaksh and Rudy
1978). This zone overlaps considerably with the region which supports SPA in the rat
(Mayer and Price 1976; Mayer and Liebeskind 1974). This region receives spinothalamic
and spino-reticulo-thalamic Somatosensory inputs, and probably contributes to the affective
components of pain.
Ventrobasal hypothalamus
Retrograde tracer injections labeled neurons in the ventrolateral, lateral, and dorsal
PAG, but few neurons were labeled in the DRN. This observation is consistent with a
previous study in the cat (Smith and Flynn 1980).
The arcuate nucleus, in the ventrobasal hypothalamus (VBH), is the major source
of 3-endorphin neurons in the brain (Finley et al. 1981). Lesions of the VBH that deplete
[3-endorphin in the brain (Millan et al. 1984), lower nociceptive thresholds, reduce stress
induced analgesia, and block PAG stimulation-produced analgesia (Kelsey et al. 1986;
Millan et al. 1983; Millan et al. 1986; Vidal and Jacob 1980). Conversely, electrical
stimulation of the VBH produces analgesia (Goodman and Holcombe 1976; Rhodes and
Liebeskind 1978; however, see Carstens 1986). The VBH projection to the PAG, which
includes 3-endorphin, enkephalin, and excitatory amino acid components (Beart et al.
1988; Bloom et al. 1978; Yamano et al. 1986) is the largest descending input to the PAG
 67

(Beitz 1982; Marchand and Hagino 1983; Meller and Dennis 1986). Conceivably,
activation of the VBH produces analgesia by initiating descending controls via the PAG. In
fact, electrical stimulation of the VBH produces naloxone-reversible inhibition of PAG
neurons (Strahlendorf et al. 1982) and excitation of neurons that project from the PAG to
the NRM (Sakuma and Pfaff 1980). Furthermore, stimulation of the VBH excites NRM
spinal projection neurons in rat (Lumb and Morrison 1986), and inhibits spinal dorsal horn
nociceptive neurons in rat and cat (Carstens 1982; Carstens 1986; Mokha et al. 1987).
Our observation that many PAG-NRM projection neurons send axon collaterals to
the VBH suggests that, in addition to activating descending controls from the PAG, the
VBH might, itself, be the target of nociception-related controls. In fact, the activity of
many VBH neurons is modulated by systemically administered morphine (Dafny 1980;
Kerr et al. 1974; Lakoski and Gebhart 1984). Since opiate binding sites (especially mu
and delta) are relatively scarce in the VBH (Mansour et al. 1987), the opioid effect is
probably mediated through PAG circuitry. In fact, microinjection of morphine in the PAG
inhibits VBH neurons (Lakoski and Gebhart 1984).
Since neurons in the VBH do not have obvious peripheral receptive fields (Dafny et
al. 1970; Sakuma and Pfaff 1982), PAG inhibition of VBH neurons probably does not
simply interrupt nociceptive transmission. Rather, modulation of VBH activity might
disrupt the affective reactions to noxious stimuli. A recent study by Pott et al. (1987)
supports this hypothesis. Affective defense was elicited in cats by electrical stimulation of
the VBH. Concurrent electrical stimulation in different zones of the PAG either suppressed
or facilitated the behavior. Since the suppression was blocked by naloxone microinjected
through the PAG electrode, and since microinjected D-Alaz-Mets-enkephalin also
suppressed the behavior, endogenous opioids are probably involved. In this way,
descending inhibition of spinal nociceptive transmission might be complemented by an
ascending suppression of integrated reactions to noxious stimuli.
 68

Medial thalamus

Much of the literature concerning the contribution of the medial thalamus to

nociceptive processing has not distinguished among the individual nuclei. Therefore, when
necessary, we use the term "medial thalamus" to refer to the intralaminar region that
includes parafascicularis (Pf), central medial, centrolateral, and centre médian (in large
mammals) thalamic nuclei, without implying that this is a homogeneous group of nuclei.
The intralaminar nuclei of the medial thalamus receive nociceptive inputs from the
spinal cord and from the brainstemreticular formation. Most studies emphasize the
contribution of intralaminar nuclei to aspects of nociception other than the spatial or
temporal discrimination of noxious stimuli. Through projections to "limbic" structures,
including much of the cerebral cortex, these medial thalamic sites are presumed to generate
affective components of pain sensation rather than the discriminative components.
Consistent with this idea, large lesions of the medial thalamus do not alter thresholds for
nociceptive reflexes (Bevan and Pert 1975; Delacour 1971; Finger and Frommer 1970;
Yaksh et al. 1977; Yeung et al. 1975), but may abolish escape reactions to electrical
stimulation (Mitchell and Kaelber 1967), disrupt conditioned avoidance responses
(Delacour 1971; Marburg 1973), and reduce chronic pain in human (Sano 1979; Sugita et
al. 1972; Urabe and Tsubokawa 1965). Consistent with these lesion studies,
microinjection of morphine in the medial thalamus can produce analgesia (Mayer and Price
1976; Pert and Yaksh 1974). In addition, ionophoretically applied morphine modulates the
spontaneous activity Pf neurons in rat (Reyes-Vasquez and Dafny 1982), and
iontophoretically applied D-ala-D-leu-enkephalin inhibits heat-evoked activity in the nucleus
lateralis anterior of rat thalamus (Hill and Pepper 1978).
It is of interest, however, that analgesia also results from Stimulation of medial
thalamus, in rat and monkey (Mayer and Leibeskind 1974; Mayer and Price 1976; Oleson
et al. 1980; Rhodes and Liebeskind 1978), and of centre median-Pf in humans for clinical
pain (Andy 1980; Richardson and Akil 1977; Thoden et al. 1979). A possible explanation
 69

for this somewhat paradoxical analgesic effect, is that medial thalamic SPA is an artifact of
the antidromic activation of PAG neurons that send axon collaterals to both the medial
thalamus and the NRM.

Since, electrical stimulation of the DRN inhibits spontaneous and tail pinch-evoked
neuronal activity in Pf of rat through a serotonergic pathway (Andersen and Dafny 1983;
Ishida and Kitano 1977), and since microintophoretically applied 5-HT has a similar effect
on Pfactivity (Andersen and Dafny 1982), it has been proposed that Pf neurons are subject
to antinociceptive controls from the DRN. However, these manipulations also inhibit
spontaneous activity and therefore, this inhibition might not be specifically antinociceptive.
We found that only a minority of PAG axons that project to medial thalamus are
serotonergic; thus, the effects of PAG/DR stimulation are probably mediated predominantly
by non-serotonergic mechanisms. In fact, Andersen (1986) found that stimulation in and
around the PAG produces IPSPs in Pf neurons that suppress footshock-induced EPSPs. It
was not determined if this effect was associated with behavioral antinociception.

TARGETS OF THE DORSAL RAPHE NUCLEUS

Consistent with previous reports, we found that neurons in DRN project to the
lateral thalamus, lateral hypothalamus, amygdala, frontal cortex, sensorimotor cortex, and
visual cortex (Azmitia and Segal 1978; Conrad et al. 1974; Moore et al. 1978); PAG.
neurons outside the boundaries of the DRN were not labeled by tracer injections into those
regions. Between 80 and 90% of DRN retrogradely labeled from these injection sites were
5-HT immunoreactive; since only 25-50% of DRN neurons are serotonergic (Moore 1981;
Descarries et al. 1982), a greater proportion of serotonergic, than non-serotonergic,
neuurons contribute to these rostral projections. -

Consistent with previous studies, DRN neurons retrogradely labeled from the RVM
injections were most common in the lateral DRN (Beitz 1982; Fardin et al. 1984a). In
contrast, neurons retrogradely labeled by forebrain injections are, for the most part, located
 7 O

more rostrally and more medially. Although different patterns of labeled neurons were
produced in the DRN by the different forebrain injections, a regular "encephalotopic"
organization of the DRN (Imai et al. 1986; Waterhouse et al. 1986) was not apparent.
Typically, axons originating in the DRN are highly collateralized (cf. DeOlmos and
Heimer 1980); our results indicate that at least 8-17% of DRN neurons that project to the
forebrain regions studied, have axon collaterals that terminate in the RVM. Fallon and
Loughlin (1982) found that neurons in the medial part of the DRN had more highly
collateralized axons (terminating in forebrain structures) than did neurons in the lateral
DRN. We did not find that double-labeled neurons were more common medially, perhaps
because most neurons that project to the NRM are located in the lateral DRN.
It has been suggested that the DRN contributes to SPA elicited from sites in and
near the PAG. For example, Fardin et al. (1984b) found that analgesia, unaccompanied by
aversive side effects, could be elicited from the dorsal part of the DRN, a region that gives
rise to an extensive system of ascending serotonergic projections. If ascending projections
from the DRN are involved in antinociceptive processes, it seems likely that collaterals
from the PAG-NRM would be involved. However, as discussed below, our data along
with a variety of other studies, provide little evidence for a specific antinociceptive function
of the DRN ascending projections.
Lateral Thalamus

Our observation that the lateral thalamus receives projections only from the DRN
and not from the PAG itself is consistent with other studies in the rat (Mantyh 1983) and
cat (Chi 1970; Hamilton and Skultety 1970), but differs from the observation by Barbaresi
et al. (1982) of a large projection from the lateral PAG to ventrobasal complex in cat.
Electrical Stimulation of the ventrobasal thalamic nuclei has been used to relieve

clinical pain in humans (Dieckmann and Witzmann 1982; Hosobuchi et al. 1973; Mazars et
al. 1979, Tsubokawa et al. 1982; Turnbull et al. 1980; Young et al. 1985). Stimulation of
the ventral posterior lateral nucleus (VPL) excites medullary raphespinal neurons in cat
 7 1

(Tsubokawa et al. 1981) and monkey (Willis et al. 1984), and inhibits spinothalamic tract
(STT) neurons (Gerhart et al. 1983). This spinal inhibition could result from antidromic
activation of STT neurons that have collaterals in the RVM or PAG (Kevetter and Willis
1983; Harmann et al. 1988), or from antidromic activation of neurons we observed in the
DRN that have collaterals in the RVM. In fact. we found that the cell bodies of neurons

retrogradely labeled from VPL are located in the dorsal part of the nucleus -- the region
from which Fardin et al., (1984b) elicited "pure" analgesia.
Since the VPL has no known descending projections, analgesic effects of electrical
stimulation in the lateral thalamus stimulation might reflect the activation of intrathalamic or
cortical projections of the ventrobasal complex (c.f. Benabid 1983). In fact, there are
several differences in the character of analgesia elicited from the PAG and from the lateral
thalamus in human (reviewed by Fields 1987). Another explanation for human lateral
thalamic SPA, based on our data in rat, is that electrical stimulation in the lateral thalamus
antidromically activates collaterals of the projection from the DRN to RVM. Such
explanations, based on anatomical studies in animals, must be tentative since electrical
stimulation of the lateral thalamus fails to produce analgesia in rat (Mayer and Liebeskind
1974; Morgan and Franklin 1988) or in monkey (Goodman and Holcombe 1976; Oleson et
al. 1980; Schmidek et al. 1971).
Lateral Hypothalamus
Electrical stimulation of the lateral hypothalamus can produce analgesia in the rat
(Balagura and Ralph 1973; Behbehani et al. 1988; however, see Mayer and Liebeskind
1974) and is antiaversive in rat (Cox and Valenstein 1965). Analgesia produced by
electrical stimulation in the lateral hypothalamic is reduced by systemically administered
opiate antagonist (Carr & Uysal 1985), by micronjection of a glutamate antagonist, or
serotonin antagonist in the NRM (Aimone and Gebhart 1988). These observations
implicate descending pathways in lateral hypothalamic SPA. However, since the lateral
hypothalamus is traversed by the medial forebrain bundle (MFB), analgesic effects of
 72

electrical stimulation in this region are particularly difficult to interpret. The MFB not only
contains afferents and efferents of the lateral hypothalamus, but also of cerebral cortex, VB
hypothalamus, parafascicularis, PAG, DRN, and NRM (Nieuwenhuys et al. 1982).
Therefore, by activating fibers of passage in the MFB, electrical stimulation in lateral
hypothalamus might activate any of these systems, including the collateral pathways we
observed in this study. In fact, Aimone and Gebhart (1987) were able to elicit analgesia by
electrical stimulation in the lateral hypothalamus, but not by microinjection of glutamate
(however, see Behbehani et al. 1988).
Telencephalon
In humans, the affective-motivational aspect of pain sensation is apparently
generated, in part, by telencephalic structures. Thus, in some chronic pain patients, frontal
lobotomy or amygdalectomy eliminates the objectional qualities of pain without reducing
the sensation of noxious stimuli (Barber 1959; Freeman and Watts 1950; Hebben et al.
1985)). Some of these patients, in fact, have a lowered threshold for pain sensation. This
latter effect is more readily observed in animals: functional lesion of the frontal cortex in
rat, or amygdala in monkey, reduces the threshold for nociceptive responses (Bagshaw and
Pribram 1968; Cooper 1976). Conversely, electrical stimulation of frontal cortex in rat, or
amygdala in monkey, can produce antinociception (Chin et al. 1976; Hardy 1985). This
effect could be produced by activation of descending inhibitory systems via orthodromic
activation of projections from the cortex to the PAG, or antidroimic activation of collaterals
from the PAG-NRM pathway. The ability of somatosensory cortex stimulation to
modulate spinal sensory transmission is well-known (Hagbarth and Kerr 1954; Lundberg
et al. 1963). However, since such cortical stimulation does not specifically inhibit
noxious-evoked activity in STT neurons (Coulter et al. 1974), it cannot be concluded that
such descending inhibition is antinociceptive.
On the other hand, telencephalic structures might be the target of antinociceptive
controls. Conceivably, the ascending collateral arm of the DRN-NRM pathway produces
 7 3

antinociception. Wang and Aghajanian (1977) showed that electrical stimulation in the
DRN does, in fact, inhibit activity of neurons in the amygdala. Since only spontaneous
activity was studied, however, this inhibition may not have been specifically
antinociceptive. Pucilowski et al. (1985) injected 5-HT into the amygdala and observed a
suppression of agressive behavior, but not of pain sensitivity. Thus, the relevance of this
ascending pathway to antinociception is dubious.
To help further assess the possibility that ascending projections from the DRN are
relevant to nociception, we made an injection of tracer into the occipital cerebral cortex. If
collateral projections from the DRN-NRM pathway were specifically concerned with
nociceptive processing, then we would expect that they would terminate only in regions
that have been implicated in nociceptive processing. Presumably, the area of visual cortex
we injected is not involved in nociception or antinociception. The results we obtained were
very similar to the results of other cortical injections. Thus, despite the fact that these
ascending projections from the DRN have collaterals that terminate in the RVM, we cannot
conclude that they are involved in any function related specifically to nociception. Instead,
it may be that these DRN projections mediate a less specific inhibition of sensory
transmission. For instance, serotonergic projections from the DRN inhibit the transmission
of visual information from the lateral geniculate nucleus to the cerebral cortex (Yoshida et
al. 1984).
Conclusion

These experiments demonstrate that the projection from the PAG to RVM contains a
significant number of axons that send collaterals rostrally. Thus, electrical stimuation of
the neurons that project from the PAG to the NRM necessarily activates a variety of
ascending collaterals. Those collaterals from the DRN-RVM projection are part of a
divergent system of serotonergic fibers that probably mediate a rather general modulation of
activity in the brain. On the other hand, in light of previous anatomical, pharmacological,
physiological, and behavioral investigations, it seems that the ascending collaterals from the
 74

PAG-RVM projection to ventrobasal hypothalamus and medial thalamus may be involved

in modulating nociceptive transmission or in mediating some other nociceptive reactions.
These projections probably contribute to a spectrum of effects elicited by any
antinociceptive manipulation in the PAG.
 75

REFERENCES

Abols, I.A. and A.I. Basbaum (1981) Afferent connections of the rostral medulla of the
cat: a neural substrate for midbrain-medullary interactions in the modulation of pain.
J. Comp. Neurol. 201: 285-297.
Aimone, L.D. and G. F. Gebhart (1987) Spinal monoamine mediation of stimulation
produced antinociception from the lateral hypothalamus. Brain Res.403: 290-300.
Aimone, L.D. and G.F. Gebhart (1988) Serotonin and/or an excitatory amino acid in the
medial medulla mediates stimulation-produced antinociception from the lateral
hypothalamus in the rat. Brain Res.450: 170-180.
Andersen, E. (1986) Periaqueductal gray and cerebral cortex modulate responses of medial
thalamic neurons to noxious stimulation. Brain Res. 375:30–36.

Andersen, E. and N. Dafny (1982) Microiontophoretically applied 5-HT reduces responses

to noxious stimuli in the thalamus. Brain Res. 241: 176-178.

Andersen, E and N. Dafny (1983) Dorsal raphe stimulation reduces responses of

parafascicular neurons to noxious stimulation. Pain 15:323-331.
Andy, O.J. (1980) Parafascicular-center median nuclei stimulation for intractable pain and
dyskinesia (painful-dyskinesia). Appl. Neurophysiol. 43: 133-144.
Azmitia, E.C. and M. Segal (1978) An autoradiographic analysis of the differential
ascending projections of the dorsal raphe and median raphe nuclei in the rat. J.
Comp. Neurol. 179: 641-668.
Bagshaw, M.H. and J.D. Pribram (1968) Effect of amygdalectomy on stimulus threshold
of the monkey. Exp. Neurol. 20: 197-202.
Balagura, S. and T. Ralph (1973) The analgesic effect of electrical stimulation of the
diencephalon and mesencephalon. Brain Res. 60: 369-379.
Barbaresi, P., F. Conti and T. Manzoni (1982) Periaqueductal gray projection to the
ventrobasal complex in the cat: a horseradish peroxidase study. Neurosci. Lett. 30:
205-209.

Barber, T.X. (1959) Toward a theory of pain: relief of chronic pain by prefrontal
leucotomy, opiates, placebos, and hypnosis. Psychol. Bull. 56: 430-460.
Basbaum, A.I. and H.L. Fields (1984) Endogenous pain control systems: brainstem spinal
pathways and endorphin circuitry. Ann Rev. Neurosci. 7:309-338.
Basbaum, A.I. and D. Menetrey (1987) Wheat germ agglutinin-apo■■RP gold: a new
retrograde tracer for light- and electron-microscopic single- and double-label
studies. J. Comp. Neurol. 261: 306–318.
Beart, P.M., L.S. Nicolopoulos, D.C. West and P.M. Headley (1988) An excitatory
amino acid projection from ventromedial hypothalamus to periaqueductal gray in the
 76

rat: autoradiographic and electrophysiological evidence. Neurosci. Lett. 85: 205

211.

Behbehani, M.M., M.R. Park and M.E. Clement (1988) Interactions between the lateral
hypothalamus and the periaqueductal gray. J. Neurosci. 8: 2780-2787.
Beitz, A.J. (1982) The organization of afferent projections to the midbrain periaqueductal
gray of the rat. Neurosci. 7: 133-159.
Beitz, A.J., M.A. Mullet and L.L. Weiner (1983) The periaqueductal gray projections to
the rat Spinal trigeminal, raphe magnus, gigantocellular pars alpha and
paragigantocellular nuclei arise from separate neurons. Brain Res. 288: 307-314.
Benabid, A.L., S.J. Henrickson, J.F. McGinty and F.E. Bloom (1983) Thalamic nucleus
ventro-postero-lateralis inhibits nucleus parafascicularis response to noxious stimuli
through a non-opioid pathway. Brain Res. 280: 217-231.
Bevan, T. and A. Pert (1975) The effect of midbrain and diencephalic lesions on
nociception and morphine induced antinociception in the rat. Fedn. Proc. Fedn.
Am. Soc. Exp. Biol. 34: 713.
Beyer, C., J.S. Tindal and C.H. Sawyer (1962) Electrophysiological study of projections
from mesencephalic central gray matter to forebrain in the rabbit. Exp. Neurol. 6:
435-450.

Bloom, F., E. Battenberg, J. Rossier, N. Ling and R. Guillemin (1978) Neurons

containing 3-endorphin in rat brain exist separately from those containing
enkephalin: immunocytochemical studies. Proc. Natl. Acad. Sci. 75: 1591-1595.
Carlton, S.M., G.R. Leichnetz, E.G. Young and D.J. Mayer (1983) Supramedullary
afferents of the nucleus raphe magnus in the rat: a study using the transcannula
HRP gel and autoradiographic techniques. J. Comp. Neurol. 214:43-58.
Carr, K.D. and S. Uysal (1985) Evidence of a supraspinal opioid analgesic mechanism
engaged by lateral hypothalamic electrical stimulation. Brain Res. 335:55-62.
Carstens, E. (1982) Inhibition of spinal dorsal horn neuronal responses to noxious skin
heating by medial hypothalamic stimulation in the cat. J. Neurophysiol. 48: 808
822.

Carstens, E. (1986) Hypothalamic inhibition of rat dorsal horn neuronal responses to

noxious skin heating. Pain 25: 95-107.
Chi, C.C. (1970) An experimental silver study of the ascending projections of the central
gray substance and adjacent tegmentum in the rat with observations in the cat. J.
Comp. Neurol. 139: 259-272.
Chin, J.H., Pribram, K.H., Drake, K. and L. Greene (1976) Disruption of temperature
discrimination during limbic forebrain stimulation in monkeys. Neuropsychologia
14:293-310.
 7 7

Chung, J.M., Kevetter, G.A., Yezierski, R.P., Haber, L.H., Martin, R.F. and W.D.
Willis (1983) Midbrain nuclei projecting to the medial medulla oblongata in the
monkey. J. Comp. Neurol. 214:93-102.
Conrad, L.C.A., Leonard, C.M. and D.W. Pfaff (1974) Connections of the median and
dorsal raphe nuclei in the rat: an autoradiographic and degeneration study. J. Comp.
Neurol. 156: 179-206.

Cooper, S.J. (1975) Anaesthetisation of prefrontal cortex and response to noxious

stimulation. Nature 254:439-440.

Cox, V.C. and E.S. Valenstein (1965) Attenuation of aversive properties of peripheral
shock by hypothalamic stimulation. Science 149: 323-325.
Coulter, J.D., Maunz, R.A. and W.D. Willis (1974) Effects of stimulation of sensorimotor
cortex on primate spinothalamic neurons. Brain Res.65: 351-356.
Culhane, E.S. and E. Carstens (1988) Medial hypothalamic stimulation suppresses
nociceptive spinal dorsal horn neurons but not the tail-flick reflex in the rat.
Brain Res. 438: 137-144.

Dafny, N. (1980) Multiunit recording from medial basal hypothalamus following acute and
chronic morphine treatment. Brain Res. 190:584-592.
Dafny, N., E. Peritz, B. Fischler and S. Feldman (1970) Patterns of firing and factors
determining responsiveness of units in the ventromedial hypothalamus in cats
following sensory stimuli. Arch. Internat. Physiol. Biochim. 78: 869-882.
Deickmann, G. and A. Witzmann (1982) Initial and long-term results of deep brain
stimulation for chronic intractable pain. Appl. Neurophysiol. 45: 167-172.
Delacour, J. (1971) Effects of medial thalamic lesions in the rat a review and interpretation.
Neuropsychologia 9:157-174.
De Olmos, J. and L. Heimer (1980) Double and triple labeling of neurons with fluorescent
substances; the study of collateral pathways in the ascending raphe system.
Neurosci. Lett. 19: 7-12.

Descarries, L., Watkins, K.C., Garcia, S. and A. Beaudet (1982) The serotonin neurons
in nucleus raphe dorsalis in adult rat: a light and electron microscope
radioautographic study. J. Comp. Neurol. 207: 239-254.
Eberhart, J.A., Morrell, J.I., Krieger, M.S. and D.W. Pfaff (1985) An autoradiographic
study of projections from the midbrain central gray, and from the region lateral to it.
J. Comp. Neurol. 241: 285-310.
Fallon, J.H. and S.E. Loughlin (1982) Monoamine innervation of the forebrain: collateral
ization. Brain Res. Bull. 9: 295-307.

Fardin, V., J.L. Oliveras and J.M. Besson (1984a) Projections from the periaqueductal
gray matter to the B3 cellular area (nucleus raphe magnus and nucleus reticularis
paragigantocellularis) as revealed by the retrograde transport of horseradish
peroxidase in the rat. J. Comp. Neurol. 223: 483-500.
 78

Fardin, V., J.L. Oliveras and J.M. Besson (1984b) A reinvestigation of the analgesic
effects induced by stimulation of the periaqueductal gray matter in the rat. II.
Differential characteristics of the analgesia induced by ventral and dorsal PAG
stimulation. Brain Res. 306: 125-139.

Feldman, S., B. Kreisel and N. Conforti (1976) Electrophysiological connections of the rat
mediobasal hypothalamus with brain areas mediating adrenocortical responses.
Brain Res. Bull. 1: 523-528.

Fields, H.L. and A.I. Basbaum (1978) Brainstem control of spinal pain transmission
neurons. Ann. Rev. Physiol. 40: 193-221.
Fields, H.L. (1987) Pain. New York: McGraw-Hill, p.321.
Finger, S. and D. Frommer (1970) Effects of cortical and thalamic lesions on temperature
discrimination and responsiveness to foot shock in the rat. Brain Res. 24: 69–89.
Finley, J.C.W., Lindström, P. and P. Petrusz (1981) Immunocytochemical localization of
3-endorphin-containing neurons in the rat brain. Neuroendocrinol. 33:28-42.
Freeman, W. and J.W. Watts (1950) Psychosurgery -- In the Treatment of Mental
Disorders and Intractable Pain (2nd ed.) Springfield, IL: Charles C. Thomas.
Gallager, D.W. and A. Pert (1978) Afferents to brain stem nuclei (brain stem raphe,
nucleus reticularis pontis caudalis and nucleus gigantocellularis) in the rat as
demonstrated by microiontophoretically applied horseradish peroxidase. Brain Res.
144: 257-275.

Gerhart, K.D., R.P. Yezierski and W.D. Willis (1983) Inhibition of primate spinothalamic
tract neurons by stimulation in ventral posterior lateral (VPLc) thalamic nucleus:
possible mechanisms. J. Neurophysiol. 49: 406-423.
Goodman, S.J. and V. Holcombe (1976) Selective and prolonged analgesia in monkey
resulting from brain stimulation. In: Advances in Pain Research and Therapy, Vol.
1. J.J. Bonica and D. Able-Fessard (eds.), New York: Raven, pp. 495-502.
Hagbarth, K.E. and D.I.B. Kerr (1954) Central influences on spinal afferent conduction.
J. Neurophysiol. 17:295-307.
Hamilton, B.L. and F.M. Skultety (1970) Efferent connections of the periaqueductal gray
matter in the cat. J. Comp. Neurol. 139:105-114.
Hardy, S.G.P. (1985) Analgesia elicited by prefrontal stimulation. Brain Res. 339; 281
284.

Harmann, P.A., S.M. Carlton and W.D. Willis (1988) Collaterals of spinothalamic tract
cells to the periaqueductal gray: a fluorescent double-labeling study in the rat. Brain
ReS. 441: 87-97.

Hebben, N., Corkin, S., Eichenbaum, H. and K. Shedlack (1985) Diminished ability to
interpret and report internal satates after bilateral medial temporal resection: case
H.M. Behav. NeuroSci. 99: 1031-1039.
 7 9

Herz, A., K. Albus, J. Metys, P. Schubert and H. Teschemacher (1970) On the central
sites for the antinociceptive action of morphine and fentanyl. Neuropharmacol. 9.
539-551.

Hill, R.G. and C.M. Pepper (1978) The depression of thalamic nociceptive neurones by D
ala2,D-leu5-enkephalin. Euro. J. Pharmacol. 47; 223-225.
Hosobuchi, Y., J.E. Adams and B. Rutkin (1973) Chronic thalamic stimulation for the
control of facial anaesthesia dolorosa. Arch. Neurol. 29:158-161.

Imai, H., Steindler, D.A. and Kitai, S.T. (1986) The organization of divergent axonal
projections from the midbrain raphe nuclei in the rat. J. Comp. Neurol. 243: 363
380.

Ishida, Y. and K. Kitano (1977) Raphe induced inhibition of intralaminar thalamic unitary
activities and its blockade by para-chlorophenylalanine in cats. Naunyn
Schmiedeberg's Arch. Pharmacol. 301: 1-4.
Kawakami, M., F. Kimura and S. Kawagoe (1976) Cholinergic and serotonergic neural
links and the inhibitory effects of hippocampus, lateral amygdala and central gray
matter on gonadotropin release. Endocrinol. Japon. 23:11-21.
Kelsey, J.E., W.A. Hoerman, L.D. Kimball, L.S. Radack and M.V. Carter (1986)
Arcuate nucleus lesions reduce stress-induced analgesia (SIA) and enhance non
opioid SIA in rats. Brain Res. 382: 278-290.
Kerr, F.W.L., J.N. Triplett and G.W. Beeler (1974) Reciprocal (push-pull) effects of
morphine on single units in the ventromedian and lateral hypothalamus and
influences on other nuclei: with comment on methadone effects during withdrawl
from morphine. Brain Res. 74: 81-103.
Kevetter, G.A. and W.D. Willis (1983) Collaterals of spinothalamic cells in the rat. J.
Comp. Neurol. 215:453-464.
Lakoski, J.M. and G.F. Gebhart (1984) Depression of neuronal activity in the
hypothalamic ventromedial nucleus of the rat following microinjection of morphine
in the amygdala or the periaqueductal gray. Neurosci. 12:255-266.
Liu, R.P.C. (1986) Spinal neuronal collaterals to the intralaminar thalamic nuclei and
periaqueductal grey. Brain Res. 365: 145-150.
Lumb, B.M. and J.F.B. Morrison (1986) Electrophysiological evidence for an excitatory
projection from ventromedial forebrain structures on to raphe- and reticulo-spinal
neurones in the rat. Brain Res. 380: 162-166.

Lundberg, A., Norrsell, U. and P. Voorhoeve (1963) Effects from the sensorimotor cortex
on ascending spinal pathways. Acta Physiol. Scand. 59:462-473.
Mantyh, P.W. (1983) Connections of midbrain periaqueductal gray in the monkey. I.
Ascending efferent projections. J. Neurophysiol. 49: 567-581.
 80

Mansour, A., H. Khachaturian, M.E. Lewis, H. Akil and S.J. Watson (1987)
Autoradiographic differentiation of mu, delta, and kappa opioid receptors in the rat
forebrain and midbrain. J. Neurosci. 7: 2445–2464.

Marburg, D.J. (1973) The effect on reaction to painful stimuli of lesions in the
centromedian nucleus in the thalamus of the monkey. Int. J. Neurosci. 9%: 153-158.
Marchand, J.E. and N. Hagino (1983) Afferents to the periaqueductal gray in the rat. A
horseradish peroxidase study. Neurosci. 9: 95-106.
Mayer, D.J. and J.C. Liebeskind (1974) Pain reduction by focal electrical stimulation of
the brain: an anatomical and behavioral analysis. Brain Res.68: 73-93.
Mayer, D.J. and D.D. Price (1976) Central nervous system mechanisms of analgesia. Pain
2: 379-404.

Mazars, G.J., L. Merienne and C. Cioloca (1979) Comparative study of electrical

stimulation of posterior thalamic nuclei, periaqueductal gray and other midline
mesencephalic structures in man. In: Advances in Pain Research and Therapy, Vol.
3, J.J. Bonica, J.C. Liebeskind and D. Albe-Fessard (eds.) Newyork: Raven pp.
541-546.

Meller, S.T. and B.J. Dennis (1986) Afferent projections to the periaqueductal gray in the
rabbit. NeuroSci. 19: 927–964.

Millan, M.H., M. J. Millan and R. Przewlocki (1984) Lesions of the hypothalamic arcuate
nucleus modify discrtete brain and pituitary pools of dynorphin in addition to 3
endorphin in the rat. Neurosci. Lett. 48: 149-154.
Millan, M.H., M. J. Millan and A. Herz (1986) Depletion of central 3-endorphin blocks
midbrain stimulation produced analgesia in the freely-moving rat. Neurosci. 18:
641-649.

Millan, M.J., R. Przewlocki, M.H. Milan and A. Herz (1983) Evidence for a role of the
ventro-medial posterior hypothalamus in nociceptive processes in the rat.
Pharmacol. Biochem. Behav. 18:901-907.

Mitchell, C.L. and W.W. Kaelber (1967) Unilateral vs. bilateral medial thalamic lesions
and reactivity to noxious stimuli. Arch. Neurol. 17:653-660.
Mokha, S.S., G.E. Goldsmith, R.F. Hellon and R. Puri (1987) Hypothalamic control of
nocireceptive and other neurones in the marginal layer of the dorsal horn of the
medulla (trigeminal nucleus caudalis) in the rat. Exp. Brain Res.65:427-436.
Moore, R.Y. (1981) The anatomy of central serotonin neuron systems in the brain. In;
B.L. Jacobs and A. Gelperin (Eds.): Serotonin Neurotransmission and Behavior,
MIT Press (Cambridge, MA) pp. 35-71.
Moore, R.Y., Halaris, A.E. and B.E. Jones (1978) Serotonin neurons of the midbrain
raphe; ascending projections. J. Comp. Neurol. 180:417-438.
 8 1

Morgan, M.J. and K.B.J. Franklin (1988) Stimulation-produced analgesia (SPA) from
brain-stem and diencephalic sites in the rat: relationships between analgesia,
aversion, seizures and catalepsy. Pain 33: 109-121.
Nashold, B.S., W.P. Wilson and D.G. Slaughter (1969) Sensations evoked by stimulation
of the midbrain of man. J. Neurosurg. 30: 14-24.
Nieuwenhuys, R., L.M.G. Geeraedts and J.G. Veenig (1982) The medial forbrain bundle
of the rat. I. General introduction. J. Comp. Neurol. 206: 49-81.
Oleson, T.D., D.B. Kirkpatrick and S.J. Goodman (1980) Elevation of pain threshold to
tooth shock by brain stimulation in primates. Brain Res. 194: 79-95.
Paxinos, G. and C. Watson (1986) The rat brain in stereotaxic coordinates. Orlando,
FL: Academic Press, Inc.

Pert, A. and T.L. Yaksh (1974) Sites of morphine induced analgesia in the primate brain:
relation to pain pathways. Brain Res. 80: 135-140.
Peschanski, M., G. Guilbaud and M. Gautron (1981) Posterior intralaminar region in rat:
neuronal responses to noxious and nonnoxious cutaneous stimuli. Exptl. Neurol.
72: 226-238.

Pittman, Q.J., H.W. Blume, R.E. Kearney and L.P. Renaud (1979) Influence of midbrain
stimulation on the excitability of neurons in the medial hypothalamus of the rat.
Brain Res. 174: 39-53.

Pott, C.B., S.Z. Kramer and A. Siegel (1987) Central gray modulation of affective defense
is differentially sensitive to naloxone. Physiol. Behav. 40: 207-213.
Pucilowski, O., Plaznik, A. and Kostowski, W. (1985) Agressive behavior inhibition by
serotonin and quipazine injected into the amygdala in the rat. Behav. Neural Biol.
43: 58–68.

Reichling, D.B. and A.I. Basbaum (1986) Some periaqueductal gray neurons which
project to the nucleus raphe magnus have ascending collaterals. Soc. Neurosci.
Abs., 12: 616.
Renaud, L.P. and J.B. Martin (1975) Electrophysiological studies of connections of
hypothalamic ventromedial nucleus neurons in the rat; evidence for a role in
neuroendocrine regulation. Brain Res. 93: 145-151.
Reyes-Vasquez C. and N. Dafny (1982) Response characteristics of thalamic neurons to
microiontophoretically applied morphine. Neuropharmacol. 21: 733–738.
Rhodes, D.L and J.C. Liebeskind (1978) Analgesia from rostral brainstem stimulation in
the rat. Brain Res. 143: 521-532.

Richardson, D.E. and H. Akil (1977) Pain reduction by electgrical brain stimulation in
man. Part 1: Acute administration in periaqueductal and periventricular sites. J.
Neurosurg, 47: 178-183
 8 2

Sakuma, Y. and D.W. Pfaff (1980) Excitability of female rat central gray cells with
medullary projections. Changes produced by hypothalamic stimulation and estrogen
treatment. J. Neurophysiol. 44: 1012-1023.
Sakuma, Y. and D.W. Pfaff (1982) Properties of ventromedial hypothalamic neurons with
axons to midbrain central gray. Exp. Brain Res. 46: 292-300.
Sano, K (1979) Stereotaxic thalamolaminotomy and posteromedial hypothalamotomy for
the relief of intractable pain. In J.J. Bonica and V. Ventafridda (eds.) Advances in
Pain Research and Therapy, Vol. 2. New York: Raven Press pp. 475-498.
Schmidek, H.H., D. Fohanno, F.R. Ervin, and W.H. Sweet (1971) Pain threshold
alterations by brain stimulation in the monkey. J. Neurosurg. 35: 715-722.
Schmued, L.C. and J.H. Fallon (1986) Fluoro-Gold: a new fluorescent retrograde axonal
tracer with numerous unique properties. Brain Res. 377: 147-154.
Smith, D.A. and J.P. Flynn (1980) Afferent projections to affective attack sites in cat
hypothalamus. Brain Res. 194: 41-51.
Strahlendorf, J.C., H.K. Strahlendorf and C.D. Barnes (1982) Inhibition of
periaqueductal gray neurons by the arcuate nucleus: partial mediation by an
endorphin pathway. Exp. Brain Res. 46: 462-466.
Sugita, K., N. Mutsuga, Y. Takaoka and T. Doi (1972) Results of stereotaxic
thalamotomy for pain. Confin. Neurol. 34:265-274.
Thoden, U., M. Doerr, G. Dieckmann and J.U. Krainick (1979) Medial thalamic
permanent electrodes for pain control in man: an electrophysiological and clinical
study. Electroencephal. Clin. Neurophysiol. 47: 582-591.
Tsubokawa, T and J. Sutin (1963) Mesencephalic influence upon the hypothalamic
ventromedial nucleus. Electroenceph. Clin. Neurophysiol. 15: 804-810.
Tsubokawa, T., T. Yamamoto, Y. Katayama and N. Moriyasu (1981) Diencephalic
modulation of activities of raphe-spinal neurons in the cat. Exptl. Neurol. 74:561
572.

Tsubokawa, T., T. Yamamoto, Y. Katayama and N. Noriyasu (1982) Clinical results and
physiological basis of thalamic relay nucleus stimulation for relief of intractable pain
with morphine tolerance. Appl. Neurophysiol. 45: 143-155.
Turnbull, I.M., R. Shulman and W.B. Woodhurst (1980) Thalamic stimulation for
neuropathic pain. J. Neurosurg. 52:486-493.
Urabe, M. and T. Tsubokawa (1965) Stereotaxic thalamotomy for the relief of intractable
pain - CEM thalamotomy. Tohoku J. Exp. Med. 85:286-298.
Vidal, C. and J. Jacob (1980) The effect of medial hypothalamus lesions on pain control.
Brain ReS. 199:89-100.

Wang, R.Y. and G.K. Aghajanian (1977) Inhibition of neurons in the amygdala by dorsal
raphe stimulation: mediation through a directr serotonergic pathway. Brain Res.
120: 85-102.
 83

Waterhouse, B.D., Mihailoff, G.A., Baack, J.C. and D.J. Woodward (1986) Topographic
distribution of dorsal and median raphe neurons projecting to motor, sensorimotor,
and visual cortical areas in the rat. J. Comp. Neurol. 249: 460-476.
Willis, W.D., K.D. Gerhart, W.S. Willcockson, R.P. Yezierski, T.K. Wilcox and C.L.
Cargill (1984) Primate raphe- and reticulospinal neurons: effects of stimulation in
periaqueductal gray or VPLc thalamic nucleus. J. Neurophysiol. 51: 467-480.
Yaksh, T.L. and T.A. Rudy (1978) Narcotic analgesics: CNS sites and mechanisms of
action as revealed by intracerebral injection techniques. Pain 4: 299-359.
Yaksh, T.L., Yeung, J.C. and T.A. Rudy (1976) Systematic examination in the rat of
brainstem sites sensitive to the direct application of morphine: observation of
differential effects within the periaqueductal gray. Brain Res. 114: 83-103.
Yaksh, T.L., Yeung, J.C. and T.A. Rudy (1977) Medial thalamic lesions in the rat: effects
on the nociceptive threshold and morphine antinociception. Neuropharmacol. 16:
107-114.

Yamano, M., S. Inagaki, S. Kito, T. Matsuzaki, Y. Shinohara and M. Tohyama (1986)

Enkephalinergic projection from the ventromedial hypothalamic nucleus to the
midbrain central gray matter in the rat: an immunocytochemical analysis. Brain Res.
398: 337-346.

Yeung, J.C., T.L. Yaksh and T.A. Rudy (1975) Effects of brain lesions on the
antinociceptive properties of morphine in rats. Clin. Exp. Pharmacol. Physiol. 2:
261-268.

Yoshida, M., Sasa, M. and S. Takaori (1984) Serotonin-mediated inhibition from dorsal
raphe nucleus of neurons in dorsal lateral geniculate and thalamic reticular nuclei.
Brain ReS. 290: 95-105.

Young, R.F., R. Kroening, W. Fulton, R.A. Feldman and I.C. Chambi (1985) Electrical
stimulation of the brain in treatment of chronic pain. J. Neurosurg. 62: 389-396.
 84

Fig. 1. This figure schematically illustrates the locations of typical injections of retrograde
tracer into the forebrain of different rats. 1. frontal cortex (lateral); 2. frontal cortex
(medial); 3. parietal cortex; 4. ventrobasal thalamus; 5. amydala; 6. ventrobasal
hypothalamus; 7. medial thalamus; 8. lateral hypothalamus; 9. occipital cortex. The
fluorescent tracer, True Blue, was injected at site l; WGAapoHRP-Au was injected at the
other sites. Numbers next to each drawing indicate the distance (in mm) rostral to
interaural zero according to the atlas of Paxinos and Watson (1986). Abbreviations: A,
amygdala; cc, corpus callosum; Cy, cingulate gyrus; CP, caudate-putamen; fr, fasciculus
retroflexus; Fr., frontal cortex; ic, internal capsule; IC inferior colliculus; mt,
mammillothalamic tract; O, occipital cortex; ox, optic chiasm; PAG, periaqueductal gray;
py, pyramidal tract; VB, ventrobasal thalamus, zi, zona incerta.
 86

Fig. 2. This figure shows a photomicrograh of an injection of the fluorescent tracer,

Fluoro-Gold in the RVM. The core of the injection, in the NRM, is outlined by
arrowheads. Abbreviations: VII, facial nucleus; Gi, nucleus reticularis gigantocellularis;
mlf, medial longitudinal fasciculus; py, pyramidal tract. Scale bar = 1 mm.
 88

Fig. 3. This bright-field photomicrograph illustrates the central core of an injection of

WGAapoHRP-Au in the medial thalamus (arrow). This injection is diagramed in Fig. 1.
Abbreviations: III, third ventricle; fr, fasciculus retroflexus; Hb, habenula; ml, medial
lemniscus; mt, mammillothalamic tract; Pf, nucleus parafascicularis of the thalamus. Scale
bar = 1mm. -
 9 ()

Fig. 4. This figure illustrates the appearance of single- and double-labeled neuronal cell
bodies after an injection of WGAapoHRP into Pf and an injection of Fluoro-gold into the
NRM. The fluorescence photomicrograph in plate A shows ventrolateral PAG neurons
retrogradely labeled with Fluoro-gold from the NRM. Plate B shows the identical field of
the section in dark field illumination. Neurons retrogradely labeled with WGAapoHRP-Au
from Pf contain refractile particles of tracer. Double-labeled neurons are indicated by
arrowheads. The photomicrograph in plate C shows neurons in the DRN retrogradely
labeled with Fluoro-gold, and plate D shows neurons retrogradely labeled with
WGAapoHRP-Au. Scale bars = 100 pum.
 92

Fig. 5. These dark-field photomicrographs show examples of the two major distributions
of retrogradely labeled neurons we observed in the PAG. In dark-field illumination,
individual WGAapoHRP-Au labeled cell bodies are distinct, even at low magnification
(arroiws). The upper plate illustrates the distribution of retrogradely labeled neurons in the
PAG after an injection into Pf. (the right side of the plate is ipsilateral to the injection site.)
Retrogradely labeled neurons are most concentrated ventrolateral to the aqueduct (AQ).
The lower figure shows neurons retrogradely labeled after a tracer injection into VB. In
this case, retrogradely labeled cell bodies are found only in the DRN.
 9 4

Fig. 6. Plate A shows 5-HT immunoreactive neurons in the lateral part of the DRN. Plate
B is a dark-field photomicrograh of the same field. The neurons shown in B have been
retrogradely labeled by an injection of WGAapoHRP-Au in Pf. Double-labeled neurons
are indicated by arrows. It is possible to identify double-labeled neurons with bright-field
illumination alone, since the immunostain is brown and the retrograde label is black, but
this is not evident in black-and-white micrographs. Scale bar = 100 p.m.
 9 6

Fig. 7. These drawings illustrate the patterns of retrogradely labeled neurons (black dots)
in the PAG after tracer injections (shown in figure 1) into regions of the brain rostral to the
PAG. The right-hand side of each drawing is ipsilateral to the injection site. Open circles
indicate neurons which were double-labeled by a tracer injection centered in the NRM. The
drawings for medial thalamus and ventrobasal hypothalamus injections represent the
number of cells labeled in a single 25 plm thick Section; all other drawings indicate the total
number of cells observed in four 25 plm sections. Numbers indicate the distance (in mm)
rostral to interaural zero according to the atlas of Paxinos and Watson (1986).
 VENTROBASAL LATERAL
HYPOTHALAMUS HYPOTHALAMUS

+ 1.7

+ 1.2

+0.7
 MEDIAL VENTROBASAL
THALAMUS THALAMUS

+0.7
FRONTAL AMYGDALA
CORTEX

+ 1.7

+1.2

+0.7
PARIETAL OCCIPITAL
CORTEX CORTEX

+1.7

+1.2

+0.7
 9 8

CHAPTER IV

The Contribution of Brainstem GABAergic Circuitry

to Descending Antinociceptive Controls.
I. GABA-Immunoreactive Projection Neurons in
the Periaqueductal Gray and Nucleus Raphe Magnus

David B. Reichling and Allan I. Basbaum

 99

SUMMARY

The fact that GABA receptor ligands influence nociceptive thresholds

when administered in the rostral ventral medulla or in the spinal cord, may
reflect the involvement of GABAergic neuronal elements in endogenous
antinociceptive pathways. The present study used immunocytochemistry and
retrograde tract tracing (using WGAapo HRP-Au) to assess the possibility that
GABAergic projection neurons contribute to the presumed antinociceptive
pathway which descends from the midbrain periaqueductal gray matter (PAG)
via the rostroventral medulla (NRM) to the spinal cord dorsal horn. In the
PAG, very few GABA immunoreactive neurons were retrogradely labeled from
the NRM. Similarly, in the NRM, very few GABA-immunoreactive neurons
were retrogradely labeled from the spinal cord. A much higher proportion of
GABA-immunoreactive neurons in the region lateral to the NRM, however,
were retrogradely labeled from the spinal cord. Eighteen percent of GABA
immunoreactive neurons were retrogradely labeled, n the nucleus reticularis
paragigantocellularis; conversely, 15% of the retrogradely labeled neurons in
this region were GABA-immunore active. These results indicate that

GABAergic projections constitute a very minor component of the PAG-NRM

spinal cord pathway; however, they may be an important component of
descending projections that originate in the medullary reticular formation
lateral to the NRM. The majority of GABAergic neurons in the PAG and NRM
are presumed to be inhibitory interneurons that directly or indirectly
regulate activity in efferent pathways from these regions.

INTRODUCTION

The inhibitory amino acid, GABA, has been implicated in

antinociceptive processses at both the brainstem and spinal cord levels. For
 1 00

example, Moreau and Fields (1986) reported that microinjection of the GABA
agonist, muscimol, into the midbrain periaqueductal gray matter (PAG)
antagonizes analgesia produced by morphine microinjected in the same site;
conversely, GABA antagonists produced analgesia. Similarly, in the
rostroventral medulla (RVM), the GABA agonist, THIP, produces hyperalgesia,
and the GABA antagonist, bicuculline, produces modest analgesia (Drower and
Hammond 1988). Finally, the GABAB agonist, baclofen, administered
intrathecally, produces potent analgesia (Wilson and Yaksh 1978).
Millhorn et al. (1987, 1898) reported that some cells in the rostroventral
medulla (RVM) that were immunoreactive for glutamic acid decarboxylase, the
GABA synthetic enzyme, could be retrogradely labeled from the spinal cord. At
Some rostrocaudal levels, more than half the GABA-immunoreactive neurons

in the RVM are also 5-HT immunoreactive. Since 85-90% of the 5-HT neurons

in the medullary nucleus raphe magnus (NRM) project to the spinal cord
(Bowker et al., submitted), these data imply that many GABAergic neurons
must project to the spinal cord.
Since the PAG contains a high density of putative GABAergic neurons
(Mugnaini and Oertel 1985; Reichling and Basbaum 1987), it is of interest to
determine if GABAergic projection neurons also contribute to the projection
from the PAG to the NRM. The present study used a combination of retrograde
tracing and GABA immunocytochemistry to identify putative GABAergic
projection neurons in the PAG. In addition, we performed a detailed analysis
of the distribution of GABA-immunoreactive projection neurons in the RVM.
A preliminary report of a portion of this work has been published previously
(Reichling and Basbaum 1987).

METHODS
 1 0 1

Six adult male Sprague-Dawley rats (Bantin Kingman) were used in this
study. The rats were anesthetized with pentobarbitol (60 mg/kg) during
surgery. Three rats received stereotaxic microinjections of the gold
conjugated retrograde tracer, WGAapo HRP-Au (Menetrey and Basbaum 1987),
into the NRM. These rats were placed in a stereotaxic head frame, the skull was
trephinated, and a 30 gauge 1 pil Hamilton syringe was used to pressure inject
0.1 pil of tracer over a period of 30 minutes. Three additional rats received
injections of the retrograde tracer into the cervical enlargement of the spinal
cord. Two injections, separated by about 2 mm rostrocaudally, were made on
each side of the cord. These four 0.1 pil injections were centered in the dorsal
horn. After recovery from anesthesia the rats displayed no obvious motor or
sensory deficits.
Seven to twenty-one days after surgery, the rats were deeply
anesthetized with pentobarbitol and perfused transcardially with 100 ml of a
fixative solution consisting of 0.05 M phosphate-buffered saline (PBS) followed
by 500 ml of 1% glutaraldehyde and 4% paraformaldehyde in 0.05 M PBS. The
brain was removed and sectioned on a Vibratome. The midbrain or medulla

was sectioned at 30 p.m. Injection sites were sectioned at 100 pum. Retrogradely
transported colloidal gold particles were made visible by a silver
enhancement reaction (Basbaum and Menetrey 1987). For GABA

immunocytochemistry the sections were first incubated for 30 minutes in a

blocking solution of 3% normal goat serum and 0.3% Triton X-100 in 0.05 M

PBS. The tissue was then incubated for 48 hours in the primary antiserum
diluted 1:3750 in 0.05 M PBS containing 0.3% Triton X-100. This antiserum
(kindly provided by Dr. Andrew Towle of Cornell University) was raised in rat
against a glutaraldehyde conjugate of GABA and bovine serum albumin (BSA).
Radioimmunoassay demonstrates that the antiserum is highly specific for
 102

glutaraldehyde fixed GABA (Lauder et al. 1986). The tissue was rinsed, and
incubated overnight in goat anti-rat IgG (1:100; Cappel). After another rinse
the tissue was incubated in rat peroxidase-anti-peroxidase (1:500; ICN
Immuno Biologicals). Finally, the peroxidase reaction was developed with
diaminobenzidine (DAB) as the chromogen. Preabsorption of the primary
antiserum with 10 || M of the GABA-BSA conjugate abolished all
immunoreactivity in the tissue sections.
In bright-field illumination, retrogradely transported WGAapo HRP-Au
appears as black punctae inside cell bodies. This retrograde label is readily
distinguished from the diffuse light brown stain associated with GABA
immunoreactivity. Thus, double-labeled neurons are easily recognized.
Sections were photographed and the distribution of single- and double
labelled neurons was drawn by camera lucida.

RESULTS

Tracer injections in the RVM were centered in the NRM; there was a
small amount of spread laterally into the nucleus reticularis
paragigantocellularis pars o (Figure 1A, for example). Tracer injections in the
spinal cord mainly involved the dorsal horn (Figure 1B, for example). Figure
2 shows photomicrographs of single- and double-labeled neurons. The WGA
apo HRP-Au retrograde label and DAB-visualized immunoreactivity can be
observed simultaneously in the light microscope, allowing unambiguous
identification of double-labeled neurons.

Periaqueductal gray matter

Our results for the PAG are presented schematically in figure 3. By
comparing the number of retrogradely labeled cell bodies to the total number
of cell bodies in Nissl-stained sections, we estimate that up to 18% of all
 1 03

neurons in the PAG were retrogradely labeled from the NRM. Retrogradely
labeled neurons were found throughout most of the PAG caudal to the
oculomotor nuclei. The only regions of the PAG where retrogradely labeled
neurons were rare were the ventromedial and dorsolateral PAG and the DRN.

At the level of the level of the oculomotor nuclei and more rostrally,
retrogradely labeled neurons were still common in the ventrolateral, lateral,
and dorsal PAG, but the dorsolateral zone, which contained few labeled cells,
enlarged. Most of the retrogradely labeled neurons were small (10-15 pum
diameter), fusiform or oval; some labeled neurons in the dorsal and ventral

parts of the PAG were larger (15-20 plm) and more multipolar. A similar

distribution of cells retrogradely labeled from the NRM has been reported in
previous studies in both the rat and cat, using a variety of tracers and
chromogens (Abols and Basbaum 1981; Beitz et al. 1983; Carlton et al. 1983;
Gallager and Pert 1978; however, see Fardin et al. 1984).
GABA-immunore active neurons comprised approximately 14% of the
total number of PAG neurons located caudal to the oculomotor nuclei. Caudal to

the oculomotor nuclei, GABA-immunoreactive neurons were common in all

regions of the PAG and overlapped extensively with the distribution of

neurons retrogradely labeled from the NRM. At more rostral levels, however,
there were marked differences between the distributions of the two

populations of neurons in the dorsal PAG. The distribution of both types of

neurons overlapped in the ventrolateral part of the rostral PAG, however
GABA-immunoreactive neurons were clustered in the dorsolateral PAG but

were not common dorsal to the aqueduct and in the lateral PAG in a pattern
almost reciprocal to the distribution of retrogradely labeled neurons in the
rostrodorsal PAG. This distribution of GABA-immunoreactive cell bodies agrees
with the results obtained by Mugnaini et al. (1985), using an anti-glutamate
 104

decarboxylase (GAD) antiserum.

Despite the intermingling of the populations of GABA-immunoreactive
neurons and retrogradely labeled neurons in most parts of the PAG, we found
few double-labeled neurons. Only 1.5% of all retrogradely labeled neurons in
the PAG were also GABA-immunoreactive, and only 2.1% of all GABA
immunoreactive neurons in the PAG were retrogradely labeled. Although it is
difficult to generalize about this small population of double-labeled neurons, it
was our impression that double-labeled neurons were predominantly located
in the ventral PAG. There was a small cluster of double labeled neurons dorsal

to the PAG, at levels caudal to the oculomotor nuclei. The morphology of

double-labeled neurons in the PAG was not distinctive.

Rostroventral medulla

Figure 4 illustrates the distribution of retrogradely labeled and

immuno-labeled neurons in the RVM in relation to the boundaries of the

reticular and raphe nuclei. These boundaries, adapted from Newman (1985)
and Paxinos and Watson (1986), are approximations that were determined from
dark-field examination of the tissue. We did not verify the delineations
cytoarchitecturally. Consistent with the results of many previous studies,
neurons retrogradely labeled from the spinal cord dorsal horn were common
the RVM, including the NRM, nucleus raphe pallidus (RPa), nucleus raphe
obscurus (ROb), nucleus reticularis paragigantocellularis (PGi), nucleus
reticularis paragigan to cellularis pars o (PGio, ), nucleus reticularis
paragigantocellularis lateralis (PGil), and nucleus reticularis gigantocellularis
(Gi; Bowker et al. 1981; Carlton et al. 1985; Kneisley et al. 1978; Kuypers and
Maiskey 1975; Tohyama et al. 1979a,b; Zemlan et al. 1984). Rostral to the
gigantocellular and paragigantocellular group of reticular nuclei, at the level
of the lateral superior olive, the more laterally situated population of
 105

retrogradely labeled cells shifts dorsally into the region of the caudal pontine
reticular nucleus.

GABA-immunore active neurons were found in the midline NRM and

RPa, and in the paragigantocellular reticular nuclei of the ventral medulla

(PGi, PGio, and PGil). GABA-immunoreactive neurons were rarer in the more
dorsally located ROb and Gi. Rostrally, at the level of the lateral superior
olivary nucleus, the distribution of GABA-immunoreactive neurons expanded
dorsally into the caudal pontine reticular nuclei. GABA-immunoreactive

neurons in the RVM were typically small (10-25 plm diameter) and multipolar.
These results are consistent with those of previous studies using GABA
immunocytochemistry (Bowker et al. 1988) and GAD immunocytochemistry
(Millhorn et al. 1987; Mugnaini and Oertel 1984).
For the most part, the populations of retrogradely labeled and GABA
immunoreactive neurons were intermingled in the RVM. The two major
exceptions are the nucleus reticularis gigantocellularis, which contains a
moderate number of retrogradely labeled neurons, but very few GABA
immunoreactive neurons, and the lateral reticular formation at the level of

the lateral superior olive, where the population of retrogradely labeled

neurons is concentrated dorsal to the population of GABA-immunoreactive
Il C u TO In S.

Nine percent of all retrogradely labeled neurons in the RVM were also
GABA-immunoreactive and 8% of all GABA-immuno reactive neurons in the

RVM were retrogradely labeled. The degree of double-labeling, however,

varied considerably among the regions studied. In the midline nuclei: NRM,
RPa, and ROb, double-labeled neurons were very rare -- only 2% of
retrogradely labeled neurons in the NRM were GABA-immunoreactive and
only 3% of GABA-immunoreactive neurons in the NRM were retrogradely
 106

labeled. In contrast, 30% of the sparsely distributed GABA-immunoreactive

neurons in the nucleus reticularis gigantocellularis were retrogradely
labeled. Other areas examined had intermediate proportions of double-labeled
neurons: in the nucleus reticularis paragigantocellularis, 15% of retrogradely
labeled neurons were GABA-immuno reactive and 18% of GABA

immunoreactive neurons were retrogradely labeled; in the nucleus reticularis

paragigantocellularis pars O., 6% of retrogradely labeled neurons were GABA
immunoreactive and 7% of GABA-immunoreactive neurons were retrogradely
labeled. Finally, we found no double-labeled neurons in the nucleus
reticularis paragigantocellularis lateralis.

DISCUSSION

The results of this study suggest that many GABAergic neurons in the
PAG are local interneurons -- although 14% of neurons in the caudal PAG were
GABA-immunoreactive, only about 2% of these neurons were retrogradely
labeled from the NRM. Based on their observation that lesions of major
sources of afferents to the DRN (the dorsal pontine tegmentum, substantia
nigra, or habenula) did not alter glutamic acid decarboxylase (GAD) activity in
the ventral PAG, Belin et al. (1979) came to a similar conclusion. Although
other possible sources of GABAergic inputs to the PAG have not been
examined, it is seems likely that most GABAergic neurons in the PAG are
interneurons.

In contrast, GABAergic projection neurons in the RVM do make a

significant contribution to medullo-spinal pathways. Although the
populations of GABA-immunoreactive and retrogradely labeled neurons were
intermingled in most regions of the RVM, the fraction of double-labeled
neurons varied considerably among the different regions. Only a small
 107

number of double-labeled neurons were found in the midline raphe nuclei;

the more laterally placed reticular nuclei contained greater numbers of
double-labeled neurons.

In light of studies by Millhorn et al. (1987, 1988) we expected to find a

relatively large proportion of retrogradely labeled GABA-immunoreactive
neurons in the RVM. There are several possible explanations for the lower
proportion of GABAergic projection neurons that we observed: (1) The

presence of WGA-apo HRP-Au in a cell somehow interferes with GABA

immunostaining. This is unlikely for several reasons: first, double-labeled

cells were as intensely immunostained as were other GABA-immunoreactive

cells, and second, we have used the tracer successfully in combination with
many other antisera (c.f. Menetrey and Basbaum 1988). (2) The difference
might relate to the site of retrograde tracer injection into the spinal cord. The
estimate that 85-90% of serotonergic neurons in the RVM project to the the
spinal cord derives from studies in which large injections of tracer were made
into unspecified regions of the gray and white matter of the spinal cord
(Bowker et al. 1987, and submitted). Our relatively more focussed injections of
WGAapo HRP-Au (which is picked up minimally by fibers of passage; Basbaum
and Menetrey 1987) were limited to the dorsal horn of the cervical spinal cord.
Conceivably, the population of RVM cells in which 5-HT and GABA coexist
project preferentially to regions of the cord other than the dorsal horn, e.g. to
the interomediolateral cell column (Loewy and McKellar 1981). (3) If the
population of neurons in which 5-HT and GABA coexist includes all of the 10
15% of serotonergic neurons that apparently do not project to the cord
(Bowker et al. 1987, and submitted), then the percentage of GABAergic
projection neurons predicted from Millhorn and coworkers' data would be as
low as 28%. In fact, at mid rostrocaudal levels of the nucleus reticularis
 108

paragigantocellularis we found that approximately 30% of GABA

immunoreactive neurons were retrogradely labeled. (4) Finally, the anti
GABA antiserum which we used might have different binding specificity from
that used by Millhorn et al. (1988). Since comparable absorption controls were
performed in both cases, and since the two reported patterns of GABA
immunoreactive cell bodies are in agreement with each other, we conclude
that explanation is unlikely.
Our results indicate that the lateral region of the RVM contains the
largest number of GABA-immuno reactive spinal projection neurons.
Although emphasis is usually placed on the NRM contribution to descending
controls, the lateral RVM, in part, mediates analgesia elicited by narcotics or
electrical stimulatuion of the PAG (Azami et al. 1982; Lovick 1985; Prieto et al.

1983; Sandkuhler et al. 1982). In light of reports that intrathecal

administration of GABAB agonists produces antinociception (Wilson and Yaksh
1978), it is conceivable that GABAergic projection neurons are involved in
descending antinociceptive pathways. Since many GABA-immunoreactive
neurons in the RVM lateral to the NRM are also 5-HT-immunoreactive

(Millhorn et al. 1988), the GABA-immunoreactive projection neurons we

observed might, in fact, contribute to descending antinociceptive serotonergic
pathways.
In conclusion, we found that few GABA-immunoreactive neurons in the

PAG project to the NRM. In contrast, a relatively greater number of GABA

immunoreactive neurons in the RVM project to the spinal cord dorsal horn.
These GABAergic projection neurons were concentrated in PGi and PGio, but
were rare in the midline raphe nuclei and in the more laterally located PGil.
We conclude that GABAergic elements that contribute to the pathway
descending from the PAG via the NRM are probably interneurons. On the
 109

other hand, the parallel antinociceptive pathway that descends to the spinal
cord from the lateral medullary reticular formation may have a significant
GABAergic component.

ACKNOWLEDGEMENTS

The authors thank Ms. Katja Kohler for artistic assistance, and Ms.
Simona Ikeda and Ms. Margaret Mayes for photographic assistance. This work
was supported by NIH Public Health Service Grants NS 14627 and 21445.
 1 1 0

REFERENCES

Azami, J., Llewelyn, M.D., and M.H.T. Roberts (1982) The contribution of
nucleus reticularis paragigantocellularis and nucleus raphe magnus to
the analgesia produced by systemically administered morphine,
investigated with the microinjection technique. Pain 12: 229-46.
Basbaum, A.I. and D. Menetrey (1987) Wheat germ agglutinin-apoh RP gold: a
new retrograde tracer for light- and electron-microscopic single- and
double-label studies. J. Comp. Neurol. 261: 306–318.
Belin, M.F., Aguera, M., Tapaz, A., McRae-DeQueurce, A., Bobillier, P. and Pujol,
J.F., GABA-accumulating neurons in the nucleus raphe dorsalis and
periaqueductal gray in the rat: a biochemical and radioautographic
study, Brain Res., 170 (1979) 279-297.
Bowker, R.M. and L.C. Abbott (submitted) A quantitative re-evaluation of the
serotonergic and non-serotonergic neurons in descending spinal
projections in the rat: evidence for a major descending multiple
neurotransmitter complex. Neurosci.
Bowker, R.M., Reddy, V.K., Fung, S.J., Chan, J.Y.H. and C.D. Barnes (1987)
Serotonergic and non-serotonergic raphe neurons projecting to the
feline lumbar and cervical spinal cord: a quantitative horse radish
peroxidase-immunocytochemical study. Neurosci. Lett. 75: 31-37.
Bowker, R.M., Abbott, L.C. and R.P. Dilts (1988) Peptidergic neurons in the
nucleus raphe magnus and the nucleus gigantocellularis: their
distributions, interrelationships, and projections to the spinal cord. In
H.L. Fields and J.M. Besson (eds.) Pain Modulation, Progress in Brain
Research, Vol. 77, Elsevier: New York, pp. 95-127.
Carlton, S.M., Chung, J.M., Leonard, R.B. and W.D. Willis (1985) Funicular
trajectories of brainstem neurons projecting to the lumbar spinal cord in
the monkey (Macaca fascicularis): a retrograde labeling study. J. Comp.
Neurol. 241: 382–404.

Drower, E.J. and D.L. Hammond (1988) GABAergic modulation of nociceptive

threshold: effects of THIP and bicuculline microinjected in the ventral
medulla of the rat. Brain Res. 450: 3.16-324.

Fardin, V., Oliveras, J.L. and J.M. Besson (1984) Projections from the
periaqueductal gray matter to the B3 cellular area (nucleus raphe
magnus and nucleus reticularis paragigantocellularis) as revealed by the
retrograde transport of horseradish peroxidase in the rat. J. Comp. Neurol.
223: 483-500.

Hammond, D.L. and E.J. Drower (1984) Effects of intrathecally administered

THIP, baclofen and muscimol on nociceptive threshold. Euro. J.
Pharmacol. 103: 121 - 125.

Kneisley, L.W., Biber, M.P. and J.H. Lavail (1978) A study of the origin of
 1 11

brainstem projections to monkey spinal cord using the retrograde

transport method. Exp. Neurol. 60: 367-378.
Kuypers, H.G.J.M. and V.A. Maiskey (1975) Retrograde axonal transport of
horseradish peroxidase from the spinal cord to brain stem cell groups in
the cat. Neurosci. Lett. 1: 9-14.

Lauder, J.M., Han, V.K.M., Henderson, P., Verdoon, T. and Towle, A.C., Prenatal
on togeny of the GABAergic system in the rat brain: an
immunocytochemical study, Neurosci, 19 (1986) 465-493.
Levy, R.A. and H.K. Proudfit (1977) The analgesic action of Baclofen [3-(4-
chlorophenyl)-Y-aminobutyric acid]. J. Pharmacol. Exp. Therap. 202: 437.
Loewy, A.D. and S. Mckellar (1981) Serotonergic projections from the ventral
medulla to the intermediolateral cell column in the rat. Brain Res. 21 1:
146-152.

Lovick, T.A. (1985) Ventrolateral medullary lesions block the antinociceptive

and cardiovascular responses elicited by stimulating the dorsal
periaqueductal grey matter in rats. Pain 21: 241-252.
Lovick, T.A. and J.H. Wolstencraft (1980) Inhibition from nucleus raphe
magnus of tooth pulp responses in medial reticular neurones of the cat
can be antagonized by bicuculine. Neurosci. Lett. 19: 325-330.
Millhorn, D.E., Hökfelt, T., Seroogy, K., Oertel, W. and A.A.J. Verhofstad (1988)
Extent of colocalization of serotonin and GABA in neurons of the ventral
medulla oblongata in rat. Brain Res. 461: 169-174.
Millhorn, D.E., Hökfelt, T., Seroogy, K., Oertel, W., Verhofstad, A.A.J. and J.Y. Wu
(1987) Immunohistochemical evidence for colocalization of Y
aminobutyric acid and serotonin in neurons of the ventral medulla
oblongata projecting to the spinal cord. Brain Res. 410: 179-185.
Moreau, J.L. and Fields, H.L., Evidence for GABA involvement in midbrain
control of medullary neurons that modulate nociceptive transmission,
Brain Res., 397 (1986) 37-46.

Mugnaini, E. and W.H. Oertel (1985) An atlas of the distribution of GABAergic

neurons and terminals in the rat CNS as revealed by GAD
immunohistochemistry, In A. Björklund and T. Hökfelt, (eds.), Handbook
of Chemical Neuroanatomy, Vol. 4, GABA and Neuropeptides in the CNS.
Part I, Elsevier, Amsterdam, pp. 436-608.
Newman, D.B. (1985) Distinguishing rat brainstem nuclei by their neuronal
morphology. I. Medullary nuclei. J. Hirnforsch. 2: 187-226.
Paxinos, G. and Watson, C., The Rat Brain in Stereotaxic Coordinates, 2nd ed.,
Academic Press, Orlando, Fl: 1986.

Prieto, G.J., Cannon, J.T., and J.C. Liebeskind (1983) Nucleus raphe magnus
lesions disrupt stimulation-produced analgesia from ventral but not
dorsal midbrain areas in the rat. Brain Res. 261 : 53-57.
 1 12

Reichling, D.B. and Basbaum, A.I., Anatomical evidence that GABA influences
the projection from the periaqueductal gray matter to the nucleus raphe
magnus, Pain, Supp. 4 (1987) S40.
Sandkühkler, J. and G.F. Gebhart (1984) Relative contributions of the nucleus
raphe magnus and adjacent medullary reticular formation to the
inhibition by stimulation in the periaqueductal gray of a spinal
nociceptive reflex in the pentobarbitol-anesthetized rat. Brain Res. 305:
77-87.

Tohyama, M., Sakai, K., Salvert, D., Touret, M. and M. Jouvet (1979a) Spinal
projections from the lower brain stem in the cat as demonstrated by the
horseradish peroxidase technique. I. Origins of the reticulospinal tracts
and their funicular trajectories. Brain Res. 173: 383-403.
Tohyama, M., Sakai, K., Salvert, D., Touret, M. and M. Jouvet (1979b) Spinal
projections from the lower brain stem in the cat as demonstrated by the
horseradish peroxidase technique. II. Projections from the dorsolateral
pontine tegmentum and raphe nuclei. Brain Res. 176: 215-231.
Wilson, P.R. and T.L. Yaksh (1978) Baclofen is antinociceptive in the spinal
intrathecal space of animals. Euro. J. Pharmacol. 51: 323-330.
Zemlan, F.P., Behbehani, M.M. and R.M. Beckstead (1984) Ascending and
descending projections from nucleus reticularis magnocellularis and
nucleus reticularisd gigan to cellularis: an auto radiographic and
horseradish peroxidase study in the rat. Brain Res. 292: 207-220.
 1 13

Fig. 1. This figure shows typical injection sites of retrograde tracer. A shows a
bright-field photomicrograph of an injection of WGAapo HRP-Au centered in
the NRM. This 100 plm section contains the center of the injection site. B
shows a diagram of the cervical spinal cord. The shaded region indicates the
total extent of two nearby injections of WGAapo HRP-Au in the dorsal horn.
- - --- --
Fig. 2. This high-magnification photomicrograph illustrates the appearance
of single- and double-labeled neurons in the nucleus paragigantocellularis.
GABA-immunoreactive neurons (open arrow) contain diffuse, brown DAB
reaction product. Retrogradely labeled cell bodies and their proximal
dendrites (curved arrow) contain black, punctate particles of silver-enhanced
WGAapo HRP-Au tracer retrogradely transported from the spinal cord. The two
tracers are visible in double-labeled neurons in bright-field illumination.
Although not apparent in this black-and-white micrograph, the difference in
color between the two tracers greatly enhances the ability to identify double
labeled neurons. Scale bar = 100plm.
 1 1 7

Fig. 3. This figure shows camera lucida drawings of typical 30 plm sections
through three rostrocaudal levels of the PAG. In each pair of drawings of half
of the PAG in a single section, the left-hand figure shows the distribution of
retrogradely labeled neurons (black dots), and the right-hand figure shows
the distribution of GABA-immunoreactive neurons (black dots). Double labeled
neurons are indicated by stars. Numbers indicate the distance (in mm) rostral
to interaural zero according to the rat brain atlas of Paxinos and Watson
(1986).
 1 19

Fig. 4. This figure shows camera lucida drawings of the distribution of single
and double-labeled neurons in typical 30 plm sections of the RVM.
Retrogradely labeled neurons are indicated by open circles, GABA
immunoreactive neurons are indicated by black dots, and double labeled
neurons are indicated by black stars. Numberrs indicate the distance (in mm)
from interaural zero according to the rat brain atlas of Paxinos and Watson
(1986). Abbreviations: VII, facial nerve; Amb, ambiguus nu.; Gi, nucleus
reticularis gigantocellularis; LSO, lateral superior olivary nu.; NVII, facial
nerve; NRM, nu. raphe magnus; PGi, nu. reticularis paragigantocellularis;
PGio. , nu. reticularis paragigantocellularis pars of ; PGil, nu. reticularis
paragigantocellularis lateral is; PnC, caudal pontine reticular nu.; py,
pyramidal tract; RPa, raphe pallidus nu.; ROb, raphe obscurus nu.; SPV, spinal
trigeminal tract; SPVO, spinal trigeminal nu. oralis.
 12 1

CHAPTER V

The Contribution of Brainstem GABAergic Circuitry

to Descending Antinociceptive Controls:
II. Electron Microscopic Immunocytochemical Evidence of GABAergic
Control over the Projection from the Periaqueductal Gray
to the Nucleus Raphe Magnus in the Rat.

David B. Reichling and Allan I. Basbaum

 1 22

SUMMARY

Pharmacological, physiological, and behavioral studies suggest that

inhibitory GABAergic neurons influence the projection from the midbrain
periaqueductal gray matter (PAG) to the medullary nucleus raphe magnus
(NRM). The present study used electron microscopic immunocytochemical
techniques to examine the morphology and synaptic relationships of GABA
immunore active terminals in the ventrolateral PAG. These putative
GABAergic terminals comprise 39% of all axon terminals in the PAG. GABA
immunore active terminals contain small, clear, pleomorphic or round,
vesicles, and 46% also contain some dense-cored vesicles. In some experiments
we also used a colloidal gold-conjugated retrograde tracer to label PAG neurons
that project to the NRM. About half of the synaptic inputs onto the cell bodies
and proximal dendrites of retrogradely labeled neurons are GABA
immunoreactive; these terminals make symmetrical contacts and rarely
contain dense-cored vesicles. These putative GABAergic synapses, that
directly control activity in PAG-NRM projection neurons, might mediate the
antinociception-related effects of exogenous GABAA receptor ligands.

INTRODUCTION

Many studies have implicated the inhibitory amino acid neurotrans

mitter, GABA, in CNS mechanisms of analgesia.(Defeudis 1983).
Electrical or chemical activation of the projection from the midbrain
periaqueductal gray matter (PAG) to the medullary nucleus raphe magnus
(NRM) produces antinociception (for review see Basbaum and Fields 1984).
The opioid peptide transmitters are believed to have exclusively inhibitory
effects in the CNS (Nicoll et al. 1980). Thus, for opiates to produce analgesia in
the PAG, it has been proposed that opioidergic synapses inhibit an
 1 23

interneuron that tonically inhibits the PAG-NRM projection neuron (Yaksh et

al. 1976; Basbaum and Fields 1984). A number of pharmacological studies
suggest that the projection from the PAG to the NRM is subject to GABAergic
inhibition, and that this inhibition antagonizes the analgesic effect of opioid
receptor ligands in the PAG (c.f. Moreau and Fields 1986). Therefore, the
hypothetical inhibitory interneuron, through which opioids exert an
antinociceptive effect in the PAG, might be GABAergic. The ventrolateral
region of the PAG supports electrical stimulation-produced analgesia and, this
region contains all the individual neural elements required by the circuitry
model described above, namely, opioid peptide terminals, GABA cell bodies and
terminals, and PAG-NRM projection neurons (see Reichling et al., submitted).
The present study uses combined GABA immunocytochemistry and
retrograde labeling at the electronmicroscopic level to test the hypothesis that
neurons projecting from the ventrolateral PAG to the NRM receive GABAergic
synaptic contacts. Preliminary reports of this work have been published
previously (Reichling and Basbaum 1987; Reichling et al. 198).

METHODS

Preparation of animals
Male Sprague-Dawley rats weighing 280-320 grams were used in this
study. The animals were anesthetized with pentobarbitol (60 mg/kg, i.p.),
placed in a stereotaxic frame, and the dura was exposed by a trephination. The
retrograde tracer, WGAapo HRP-Au (0.2 pil; Basbaum and Menetrey 1987) was
stereotaxically microinjected through a 30 guage Hamilton syringe into the
NRM. After survival for 5-18 days the rats were perfused transcardially with
100 ml of 37° C 0.05 M phosphate buffered saline, pH 7.4 (PBS) followed by 500
ml of fixative (described below) at 4° C. After perfusion the brains were
 1 24

dissected out, placed in cold fixative for 2 hour, and transferred to cold 0.05 M
PBS.

Pre-embedding immunocytochemistry
For experiments in which immunocytochemistry was performed before
the sections were embedded in plastic, the fixative was 0.5% glutaraldehyde
and 4% paraformaldehyde in 0.05 M PBS. Thirty-micron Vibratome sections of
the midbrain were collected, and then the tissue underwent a silver

intensification procedure to make the retrograde tracer visible at the light

microscopic level and to facilitate its detection at the electron microscopic
level. This reaction deposits a dense precipitate of silver ions on the colloidal
gold particles. A detailed protocol has been published previously (Basbaum
and Menetrey 1987). Briefly, the sections were first rinsed for 5 minutes in
0.1M sodium citrate buffer (pH 3.8), and then placed in a physical developer
solution for 50 minutes in total darkness. Each 100 ml of developer consists of:
60 ml of 50% aqueous gum arabic, 10 ml of 1.0 M sodium citrate buffer (pH 3.8),
15 ml of 5.6% aqueous hydroquinone, and 15 ml of 0.7% aqueous silver lactate.
After development, the sections were rinsed for 5 minutes in citrate buffer,
rinsed 5 minutes in 0.05 M PBS, and then placed to a solution of 2.5% sodium
thiosulfate in 0.05 M PBS for 5 minutes. Sections were rinsed at least 3 times in

0.05 M PBS before further processing.

To protect the silver-enhanced tracer against oxidation that
occurs during osmication, the tissue next underwent a gold-toning reaction.
The sections were washed three times in distilled water, and then was

incubated, in the dark, in 0.05% aqueous chloroauric acid, followed by three

distilled water washes. The gold chloride was then reduced by a two-minute
rinse in 0.05% aqueous oxalic acid, followed by three distilled water washes.
Three twenty-minute incubations in 1% aqueous sodium thiosulfate de
 1 25

impregnated excess silver. Finally, the tissue was washed three times in
distilled water.

The sections were then immunostained according to the peroxidase anti

peroxidase (PAP) immunocytochemical technique (Sternberger 1979) using a
rat polyclonal antiserum (kindly provided by Dr. Andrew Towle of Cornell
University), that was raised against a glutaraldehyde conjugate of GABA and
bovine serum albumin (BSA). Radioimmunoassay established that this
antibody is highly specific for glutaraldehyde-fixed GABA (Lauder et al. 1986).
Tissue staining was completely abolished by preabsorption of the antiserum
with 10 plM of the GABA-BSA conjugate. To improve penetration of the
antibodies, incubation times were prolonged -- 48 hours in primary antibody
(diluted 1:3750), 12 hours in goat anti-rat IgG (1:100; Cappel), and 1 hour in rat
PAP (1:500; ICN Immunobiologicals). After the PAP step, the sections were
rinsed in PBS and then reacted with the chromogen, diaminobenzidine. Small
areas of the ventrolateral PAG were excised from the sections, osmicated (1%

osmium in 0.05 M PBS for 1 hour on ice), dehydrated, and flat-embedded in

Polybed (PolyScience).
Post-embedding immunocytochemistry
For experiments using post-embedding immunocytochemistry, the rats
were perfused with a fixative solution containing 1.5% glutaraldehyde and 2%
paraformaldehyde in 0.05 M phosphate buffered saline. Thirty-micron
Vibratome sections were silver intensified and gold-toned as described above.
Small regions of tissue were excised from the ventrolateral PAG. This tissue
was then fixed in 0.5% osmium in 0.05 M PBS for 1 hour on ice, dehydrated, and
embedded in Polybed (PolyScience), a soft plastic embedding medium that is
suitable for post-embedding immunocytochemistry. This procedure is a
modification of that described by de Biasi and Rustioni (1988). After trimming
 126

the blocks to size, ultrathin sections, were collected onto Formvar-coated 200

mesh nickel grids. For immunocytochemistry, the sections were etched in a

saturated aqueous solution of NaIO4 for 30-45 minutes to expose antigenic sites.
This was followed by a five-minute wash in NaBH4 to quench nonspecific
staining. After a 30 minute incubation in normal goat serum diluted 1:30 in a
buffer consisting of 0.1% bovine serum albumin and 0.01% sodium azide in 0.05
M Tris-buffered saline pH 7.2, the immunoreaction proceeded as follows: At
room temperature, grids were placed for 1 hour in rabbit anti-GABA IgG
(Immunonuclear) diluted 1:1000 in the buffer described above; rinsed in
buffer; incubated 1 hour in goat anti-rabbit IgG conjugated to colloidal gold
particles (10 nm; Janssen) diluted 1:25 in buffer; rinsed; and counterstained
with lead citrate and uranyl acetate.
To study the synaptic organization of GABAergic elements in tissue
undamaged by the silver intensification procedure, half of the sections for
post-embedding immunocytochemistry were neither silver reacted nor gold
toned.

Analysis
Quantitative ultrastructural data were derived from the immunogold
stained (post-embedding) tissue. To estimate the density of labeled and
unlabeled GABA-immunore active terminals in the ventrolateral PAG, a

montage was constructed of 15,000X photographs covering a 200 pm? area of

tissue; the number of gold particles in each vesicle-containing profile was

counted, and the density was calculated based on the cross-sectional area of the
terminal. For analysis of the ultrastructural morphology of GABA
immunoreactive terminals, all 134 labeled terminals within five 40 pum? a■ CaS

of the ventrolateral PAG were photographed at a magnification of 20,000X.

 127

RESULTS

Retrograde label
Figure 1 illustrates a typical injection of WGA-apoh RP-Au centered in
the NRM. To maximize the number of retrogradely labeled neurons in the PAG,
we deliberately made large injections. Only a small amount of tracer spread
laterally into the adjacent nucleus reticularis gigantocellularis pars of ,
presumably because the large size of the tracer complex limits its mobility in
the tissue. Some tracer spread dorsally along the needle track into the medial
longitudinal fasciculus.
At the light microscopic level, with bright field illumination, silver
enhanced WGA-apo HRP-Au is a black, punctate precipitate in cell bodies and
proximal dendrites (figure 2). Retrogradely labeled neurons are most
numerous in the ventrolateral and lateral regions of the PAG, comprising
about one third of the total population of neurons in this region (figure 2.1).
Retrogradely labeled neurons in the ventrolateral PAG are small (10-15 pum
long diameter), fusiform or multipolar, spherical cells. Another group of
large (15-20pm long diameter), round or polygonal, multipolar labeled
neurons was observed dorsal to the aqueduct. At the level of the inferior
colliculus, another distinctive cluster of labeled, large (15-20 plm), multipolar,
round or polygonal neurons was seen in the lateral part of the ventromedial
PAG, in a region that may correspond to the lateral part of the dorsal raphe
nucleus. Since these neurons may be a subset of dorsal raphe serotonergic
neurons, the blocks of tissue used in the present study did not include this
cluster of cells.

Although gold-toning protects most of the silver-enhanced

WGAapo HRP-Au from oxidation by osmium (which results in the loss of visible
tracer particles), much of the labeling in proximal dendrites is still lost. For
 128

this reason, all the present observations concern GABA-immunore active

inputs onto neuronal cell bodies and large proximal dendrites. In the electron
microscope, the retrograde label appears as large (up to approximately 0.3 plm
diameter) electron-dense particles. These particles are observed only within
neuronal cell bodies and, occasionally, in proximal dendrites. The particles
are almost always seen in large (approximately 0.3-0.8 plm diameter) lysosomes
(see figure 8), which otherwise have a fine granular content.
Most of the retrogradely labeled neurons received very few somatic
synapses. We estimated the extent of synaptic inputs to cell bodies by making a
linear measurement of the percent coverage of the perikaryal membrane by
synapses and closely apposed terminals. This analysis was made in the plane
of a single section. Consistent with the general scarcity of axo-somatic
synapses in the ventrolateral PAG (Moss and Basbaum 1983), we measured
coverage values of less than 5% for most neurons. Some neurons, however,
including some that were retrogradely labeled had coverage values of up to
30%.

GABA Immuno reactivity

At the light microscopic level, GABA immunoreactivity in pre
embedding, DAB-reacted tissue, appeared as a diffuse stain throughout the PAG
Individual varicosities were only discernable at high magnification (figure
2). The GABA-immunoreactive fibers are densest in an annulus around the
aqueduct and in the medial part of the ventrolateral PAG. For a complete
report of our light-microscopic observations of GABA immunoreactivity in the
PAG, see Reichling et al. (submitted).
Because of the small size of colloidal gold particles, and because cell
bodies were not labeled, post-embedding immuno-gold labeling is only visible
in the electron microscope. A low background density of gold particles
 129

(approximately 0.5 particle/um?) is distributed over the entire tissue section.

Since gold particles are found at much higher density within vesicle
containing profiles, immunolabeled profiles are readily distinguished from
background. To establish a criterion to differentiate labeled from unlabeled
profiles, we measured the densities of gold particles in 50 randomly selected
terminals in the ventrolateral PAG. The results are presented graphically in
figure 2.2. Two distinct populations are evident -- "unlabeled" profiles contain
less than 10 particles/um” and "labeled," or "immunoreactive," profiles
contain more than 25 particles/um”. Using these criteria, 133 vesicle
containing profiles were classified as GABA-immunoreactive in a 2000 um?
region of the ventrolateral PAG; 207 were unlabeled. Thus, approximately 39%
of all terminals in the ventrolateral PAG are GABA-immunoreactive.

Colloidal gold particles do not preferentially cluster over the synaptic

vesicles within immunoreactive terminals. Instead, single gold particles
and/or clusters of 3-4 particles are spread evenly within the confines of the
terminal. A higher density of gold particles was sometimes seen over
mitochondria that were located in the immunoreactive synaptic terminals.
The mean area of GABA-immunoreactive profiles is 3.7 pm2. This figure
does not differ significantly from the mean for unlabeled profiles.
Immunoreactive profiles contain electronlucent cytoplasmic matrix and
always contain small, agranular vesicles (40-50 nm diameter) that, under our
fixation conditions, are round or occasionally pleomorphic (figure 3). The

packing density of these vesicles varies widely, from sparse (40 vesicles/um”),
to very dense (250 vesicles/um”). Dense-cored vesicles are also found in 46%

of immunoreactive profiles (figures 3 and 6); in such terminals, 2-4 of these

large (80-95 nm diameter) granular vesicles are usually seen in a single plane
of section.
 130

We could identify synaptic specializations in approximately 30% of the

immuno-gold labeled profiles. A single section through the terminal usually
revealed only a single zone of synaptic contact that typically measured 0.15
0.40 plm in length. The membranes in contact were usually straight or
slightly curved. Synaptic contacts with details distinct enough to allow
classification always appeared to be Gray's Type II, i.e. with "symmetrical" pre
and postsynaptic densities (figures 3 and 4).
The postsynaptic element of most of the GABA-immunoreactive synaptic
contacts was identified morphologically as a dendrite ranging in area from 1
20 um? with a median of 2 um”. Only about 5% of the labeled terminals made
synaptic contact with cell bodies. We found no evidence for GABA

immunoreactive axo-axonic synapses in the ventrolateral PAG.

Synaptic Contacts with Retrogradely Labeled Neurons
Some GABA-immunoreactive synaptic terminals made synaptic contact
with retrogradely labeled cell bodies and proximal dendrites (figures 6 and 7).
Retrogradely labeled neuronal cell bodies that were contacted by GABA
immunoreactive terminals were small (approximately 15 x 8 pum) and oval
shaped. These neurons probably correspond to the small, bipolar fusiform or
oval, neurons seen in the light microscope. The cytoplasm of these cells had
no remarkable features. Neither the morphology of these terminals nor their
synaptic contacts distinguished them from the general population of GABA
immunoreactive terminals. These terminals, however, were distinctive with

respect to their vesicle content: with only one exception, all of the GABA
immunoreactive terminals that were presynaptic to projection neurons
contained densely-packed round or pleomorphic, small, agranular vesicles,
but none contained dense-cored vesicles.
 13.1

The necessity of performing electron microscopic studies to

unequivically establish the existence of synaptic contacts was evident in these
studies. Specifically, we occasionally observed terminals that appeared to
contact a dendritic profile, but higher magnification revealed that the apposed
profiles were, in fact, separated by a very thin sheet-like process (figure 5).
These processes, presumably astrocytic in origin, were in some cases as thin as
20 nm.

In order to estimate the relative contribution of GABAergic somatic

inputs to these projection neurons, we counted the number of GABA
immunopositive and GABA-immunonegative synaptic contacts on 18 randomly
selected, retrogradely labeled cell bodies and 3 proximal dendrites. Together,
these profiles received 11 immunopositive contacts and 10 immunonegative
synaptic contacts. Based on this small sample, it appears that putative
GABAergic terminals constitute approximately half of the observed synaptic
inputs to the cell bodies of some projection neurons in the PAG.

DISCUSSION

This study provides new information on the organization of putative

GABAergic neuronal elements in the ventrolateral PAG of the rat. We found
that 39% of axon terminals in the ventrolateral PAG were GABA

immunoreactive. These GABA-immunoreactive terminals contain numerous

small, clear vesicles, and many also contain dense-cored vesicles. Importantly,
neurons that were retrogradely labeled from the NRM received synaptic
contacts onto their cell bodies and proximal dendrites. These contacts

accounted for approximately half of the observed synaptic inputs to these

Il CurOn S. These data provide support for the hypothesis that GABAergic
interneurons control activity in PAG neurons that project to the NRM.
 132

GABA immuno reactivity

Our initial studies characterized the overall distribution of GABA

immunoreactive terminals in the PAG using the pre-embedding PAP-reacted

material. This distribution has been reported in detail previously (Reichling
et al., submitted). Briefly, GABA-immunoreactive terminals are moderately
dense throughout the PAG, but are densest in an annulus surrounding the
aqueduct and in the ventrolateral PAG.
The punctate nature of post-embedding immunogold stained material
permits recognition of even very low density staining. GABA-immunoreactive
material, may, however, result from metabolic (mitochondrial) processes in
the cell or from accumulation of extracellular GABA by neurons and glia that
do not produce GABA as a transmitter (Csillag et al. 1987; Izumi et al. 1980;
Kisvárday et al. 1986; Zucker et al. 1984). It is, therefore, important to establish
quantifiable criteria to distinguish what might be non-transmitter-ded rived
staining from specific staining of putative GABAergic synaptic terminals.
Since there is such a clear separation between the populations of densely and
lightly labeled terminals, we conclude that terminals with greater than 25
colloidal gold particles/um? are "labeled" and presumably GABAergic. The
fact that a single thin section contains heavily labeled terminals that are
immediately adjacent to unlabeled terminals (for example, see figure 3)
provides some assurance that unlabeled terminals are not "false negatives" --
we believe that these two varieties of staining are due to properties of the
terminals and are not due to vagaries of the staining process.
We observed that the anti-GABA antibody binds throughout the terminal
profile, including the mitochondria. This distribution of antigenic sites may
be related to the process of glutaraldehyde fixation, by which the amino group
of the GABA molecule is cross-linked to proteins. Proteins rich in arginine
 133

and lysine offer the most amino sites for cross-linkage, and so the staining
pattern we observed might indicate a differential distribution of available
cross-linkage sites in the terminal, rather than the distribution of GABA in
vivo. There is no reason to believe that this fixation artifact reduces the

reliability of our method to identify putative GABAergic terminals. In fact,

Schon and Iversen (1972) offered a similar explanation for the preferential
binding of [3H]GABA to neuronal nuclei.

We found that 39% of all terminals in the ventrolateral PAG are GABA

immunoreactive, a figure that is consistent with previous counts in other

regions of the brain. For example, Iversen and Bloom (1972) reported that 25
45% of synaptosomes in homogenates of rat brain accumulate [3H]GABA, and
using post-embedding immunocytochemistry, Cho and Basbaum (unpublished)
found that 45% of terminals in the NRM are GABA-immunoreactive, and

Beauvillain, et al. (1988) found that 40% of the boutons in guinea pig
hypothalamus are GABA-immunoreactive.
Since 46% of GABA-immunoreactive terminals contain dense-cored

vesicles, it is likely that coexistence with peptides is an important

characteristic of GABAergic synaptic transmission in the PAG. Coexistence of
GABA with putative peptide transmitters in the PAG has, however, not yet been
demonstrated. In light of pharmacological evidence for antinociception
related interactions between GABA and opiates in the PAG (c.f. Moreau and
Fields 1986) it is of interest that GABA coexists with enkephalin in some retinal
amacrine cells (Watt et al. 1988), where it has been suggested that enkephalin
exerts an autoregulatory influence over GABAergic transmission (Lam et al.
1986).

Retrograde Labeling
 134

All injections of retrograde tracer were centered in the NRM. WGA

apoRRP-Au is poorly taken up by axons of passage, and is not transported
anterogradely (Basbaum and Menetrey 1987). Thus, it is likely that all retro
gradely labeled neurons observed in the PAG have axon terminals in the
immediate vicinity of the injection site in the rostroventral medulla. Due to
the large size of the WGAapo HRP-Au complex, which limits its mobility in
tissue, only small amounts of tracer spread laterally from the NRM into the
nucleus reticularis gigantocellularis pars o (NG co, ), which receives a
significant projection from the ventrolateral PAG (Mantyh 1983). Consistent
with the idea that most neurons were retrogradely labeled from the NRM, the
distribution of labeled neurons we found corresponds to that reported in other
studies where different retrograde tracers were used (Abols and Basbaum
1981; Beitz et al. 1983; Carlton et al. 1983; Fardin et al. 1984a; Gallager and Pert
1978). Nevertheless, since both the NGco, and NRM have been implicated in
descending antinociceptive effects elicited from the PAG (Lovick 1985; Prieto
et al. 1983; Sandkühler et al. 1984), the following discussion bears on PAG
activation of both medullary sites.
GABA-immuno reactive synaptic contacts on to projection neurons
Background staining with WGAapo HRP-Au is almost nonexistant at the
electron microscopic level. Thus, the ultrastructural identification of

retrogradely labeled neuronal elements is unambiguous. This assertion is

supported by the following observations: 1) silver-enhanced particles of

retrograde tracer are found only in lysosomes in neuronal cell bodies and
proximal dendrites; 2) labeled profiles usually contain more than one particle
of the tracer.

We found that approximately half the somatic synapses on retrogradely

labeled neurons were GABA-immunoreactive. The strategic location of
 135

putative GABAergic synapses on cell bodies suggests that activity in some

components of the PAG-NRM pathway are subject to GABAergic gating.
Consistent with GABAergic terminals exerting an inhibitory action, all of the
labeled terminals we observed formed symmetrical Gray type II synaptic
contacts, and many contained pleomorphic vesicles. Importantly, we
occassionally observed thin glial sheaths interposed between an immuno
labeled terminal and the cytoplasmic membrane of a retrogradely labeled cell
body. The existence of these near appositions emphasizes the danger in
attaching much significance to reports of "apparent contacts" based on light
microscopic observations -- confirmation of synaptic contact can only be
made at the electronmicroscopic level.
The origin of GABA-immunore active terminals
Since GABA-immunoreactive neurons are very common in the PAG
(Clements et al. 1985; Reichling et al. 1987; Reichling et al., submitted), it is
likely that the GABA-immunoreactive terminals we observed derive locally,
from GABAergic cell bodies located within the PAG. In fact, other lines of
evidence support the idea that GABAergic neurons in the PAG have local axon
terminals. First, lesions of major sources of afferents to the dorsal raphe
nucleus (the dorsal pontine tegmentum, substantia nigra, or habenula) do not
alter glutamic acid decarboxylase (GAD) activity in the ventral PAG and dorsal
raphe of the rat (Belin et al. 1979). Second, light microscopic Golgi studies
(Tredici et al. 1983) show that PAG cell bodies often appear to be contacted by
axons that originate from nearby neurons. These investigators also reported
that all axo-somatic synapses in the PAG have symmetrical junctional
morphology and contain pleomorphic vesicles, and infered that these synaptic
inputs comprise a uniformly inhibitory population of synapses. Our data

contradict the conclusion that the population of axo-somatic synapses in the

 136

PAG is homogeneous -- some are GABA-immunoreactive and some are

unlabeled.

In other studies, Sandner et al. (1986) electrically stimulated and

recorded at nearby sites in the rat PAG. Stimulation-induced inhibition of PAG
neurons was often observed, and the latency of inhibition was consistent with
mediation by a local inhibitory synapse. The post-stimulus patterns of activity
suggested that both recurrent inhibition and local interneurons were
involved. Since these inhibitory effects were most common when the
electrodes were separated in the mediolateral, versus rostrocaudal, direction,
local inhibitory connections in the PAG might be preferentially oriented in
the coronal plane.
The contribution of GABAergic controls to antinociception
Although multiple pathways mediate antinociception elicited from the
PAG, the most thoroughly characterized system involves the direct projection
to the NRM. The latter region is the origin of an inhibitory serotonergic
projection to the spinal cord dorsal horn (see Basbaum and Fields 1984). Since
electrical stimulation in the PAG produces analgesia (for review see Mayer and

Liebeskind 1974), and since lesions of the PAG do not produce analgesia
(Dostrovsky and Deakin 1977; Kelly and Glusman 1968; Melzack et al. 1958;
Rhodes 1979; Yeung et al. 1975), it has been concluded that the projection
neuron from the PAG to the NRM must be activated for analgesia to result.
Narcotics microinjected in the PAG also produce analgesia (for review see
Yaksh and Rudy 1978) at least in part, via the same circuitry through which
electrical stimulation produces analgesia (for discussion see Besson and
Chaouch 1987, pp. 144-145). Opiates, however, are presumed to have
exclusively inhibitory effects (Nicoll et al. 1980), and thus could not directly
activate the PAG neurons that project to the NRM. It has, therefore, been
 1.37

proposed that opiates indirectly activate the PAG-NRM projection neurons, via
a disinhibition. Specifically, opiates are presumed to inhibit an inhibitory
interneuron that tonically inhibits the PAG output neuron (Yaksh et al. 1976).
Zieglgänsberger et al. (1979) reported a similar disinhibitory action in
the hippocampus that was mediated by opiate-sensitive, tonically active
GABAergic interneurons. In view of this model, and the presence of putative
GABAergic neurons in the PAG, Basbaum and Fields (1984) proposed that a
similar tonically active, GABAergic interneuron is interposed between the
opioidergic terminals in the PAG and the neurons that project to the NRM.
This model predicts the existence of two synapses in the PAG: 1) opioid
terminals that contact GABA neurons, 2) GABA terminals that contact PAG

NRM projection neurons. The present study demonstrates that GABA terminals
do, in fact, synapse upon PAG neurons that project to the rostral ventral
medulla.

Although we cannot conclude from our data that these GABAergic

elements are subject to opioidergic controls, pharmacological studies support
that hypothesis. For example, if opiates produce analgesia by reducing
GABAergic controls over the PAG-NRM projection, then enhancing
transmission at the GABAergic synapse should antagonize morphine analgesia.
Ho et al. (1976), in fact, observed that inhibiting GABA uptake, or inhibiting
the GABA-degradative enzyme, GABA transamimase, raised the ED50 for
morphine-induced antinociception in mice. Conversely, blocking GABAergic
transmission should mimic the effects of morphine; Ho et al. (1976) indeed
found that systemically administered bicuculline, a GABA antagonist, lowered
the ED50 for morphine. Other studies indicate that the GABA receptors that
mediate these effects are supraspinal. Intracerebroventricularly (i.c. v.)
injected GABA antagonizes antinociception produced by i.c. v. D-Alaº-Met
 1 38

enkephalinamide or morphine (Izumi et al. 1980; Mantegazza et al. 1979).

Conversely, bicuculline has an antinociceptive action when administered
intracisternally, but not intrathecally (Ueda 1987).
Microinjection studies indicate that the relevant GABA-sensitive
circuitry is localized within the PAG. Thus, microinjection of GABA
antagonists (bicuculline or picrotoxin) in the PAG produces antinociception
(Depaulis, et al. 1987; Moreau and Fields 1986), and microinjection of the GABA
agonist, muscimol, in the PAG antagonizes the antinociceptive effect of
systemic morphine (Romandini and Samanin 1984; Zambotti, et al. 1982).
Furthermore, since GABA agonists (muscimol or THIP) microinjected in the
PAG antagonize antinociception produced by morphine microinjected at the
same site (Depaulis, et al. 1987; Moreau and Fields 1986) it can be concluded that
the relevant opiate-sensitive circuitry, as hypothesized, is also located in the
PAG.

The same proposed circuit might also account for the antinociceptive
effects of the benzodiazepines, drugs which enhance the inhibitory effect of
GABA at the GABAA receptor complex (for review see Haefeley and Polc 1986).
The benzodiazepine antagonist and inverse agonist, Ro15-1788, produces
analgesia when administered systemically to the rat (Morgan et al. 1987). The
PAG contains dense neuronal-type benzodiazepine binding sites (Young and
Kuhar 1980). If Ro15-1788 were acting at the GABAergic synapse in the PAG
which we observed, it would release the PAG-NRM projection neuron from
(presumed) tonic GABAergic inhibitory control. This disinhibition would

mimic the proposed antinociceptive action of opiates, but since benzodiazepine

receptor ligands (and GABA receptor ligands) would act on circuitry in the
PAG downstream from opioidergic synapses, analgesia produced by the former
drugs should not be reversed by opiate antagonists in the PAG.
 139

Autoradiographic receptor binding studies showed that GABAA binding

sites are much more dense than GABAB binding sites in the PAG. This is
particularly so in the caudal PAG, where analgesic manipulations are most
effective (Bowery, et al. 1987; Yaksh et al. 1976). Since GABAA receptors
mediate post-synaptic, rather than presynaptic inhibition, it is significant
that all of the GABA-immunoreactive terminals we observed in the caudal,

ventrolateral PAG were either axo-dendritic, or, more rarely, axo-somatic; we

never observed GABA-immunore active axo-axonic synapses. These

observations suggest that actions of GABA at the synapses we observed are

mediated by GABAA receptors. In fact, all of the pharmacological studies
described above, which are consistent with an action at those synapses, used
ligands that act on the GABAA receptor complex.
On the other hand, the an algesia-producing effects of GABA receptor
agonists (Hynes et al. 1984; Vaught et al. 1985; Retz and Holaday 1986), might
reflect an action on GABAB receptors located in the spinal cord (Hammond and
Drower 1984; Sawynok 1983; Vaught et al. 1985; Wilson and Yaksh 1978) or at
supraspinal sites (Lim et al. 1985; Proudfit and Levy 1978; Zorn and Enna 1987).
In conclusion, these experiments demonstrate that a large proportion of
axon terminals in the ventrolateral PAG are GABA-immunoreactive. Some of

these putative GABAergic axon terminals make symmetrical synaptic contacts

onto neurons that project to the NRM. Thus, exogenously applied GABA
receptor ligands may produce antinociception-relasted effects by acting at
these synapses, perhaps by binding to GABAA receptors. Future studies will be
directed towards characterizing the synaptic relationships between opioid
terminals and GABAergic neurons in the PAG.

ACKNOWLEDGEMENTS
 140

The authors thank Ms. Bonnie Lord for technical assistance, Ms. Katja
Kohler for artistic assistance, and Ms. Margaret Mayes and Ms. Simona Ikeda
for photographic assistance. This work was supported by NIH Public Health
Service Grants NS 14627 and 21445.
 141

REFERENCES

Abols, I.A. and A.I. Basbaum (1981) Afferent connections of the rostral medulla
of the cat: a neural substrate for midbrain-medullary interactions in
the modulation of pain. J. Comp. Neurol. 201: 285-297.
Basbaum, A.I. and H.L. Fields (1984) Endogenous pain control systems:
brainstem spinal pathways and endorphin circuitry. Ann. Rev. Neurosci.
7: 309-338.

Basbaum, A.I. and D. Menetrey (1987) Wheat germ agglutinin-apoh RP gold: a

new retrograde tracer for light- and electron-microscopic single- and
double-label studies. J. Comp. Neurol. 261: 306–318.
Beauvillain, J.C., Mitchell, V., Tramu, G. and M. Mazzuca (1988) GABA axon
terminals in synaptic contact with enkephalin neurons in the
hypothalamus of the guinea pig. Demonstration by double
immunocytochemistry. Brain Res. 443: 315-320.
Beitz, A.J. and R.D. Shepard (1985) The midbrain periaqueductal gray in the rat.
II. A Golgi analysis. J. Comp. Neurol. 237: 460-475.
Beitz, A.J., M.A. Mullet and L.L. Weiner (1983) The periaqueductal gray
projections to the rat spinal trigeminal, raphe magnus, gigantocellular
pars alpha and paragigantocellular nuclei arise from separate neurons.
Brain Res. 288: 307-314.

Belin, M.F., Aguera, M., Tapaz, A., McRae-DeQueurce, A., Bobillier, P. and J.F.
Pujol (1979) GABA-accumulating neurons in the nucleus raphe dorsalis
and periaqueductal gray in the rat: a biochemical and radioautographic
study. Brain Res. 170: 279-297.
Besson, J.M. and A. Chaouch (1987) Peripheral and spinal mechanisms of
nociception. Physiol. Rev. 67: 67-186.
Bowery, N.G., Hudson, A.L. and G.W. Price (1987) GABAA and GABAB receptor site
distribution in the rat central nervous system. Neurosci. 20: 365-383.
Carlton, S.M., G.R. Leichnetz, E.G. Young and D.J. Mayer (1983) Supramedullary
afferents of the nucleus raphe magnus in the rat: a study using the
transcannula HRP gel and autoradiographic techniques. J. Comp.
Neurol. 214: 43-58.

Clements, J.R., Beitz, A.J., Larson, A.A. and J.E. Madl (1985) The ultrastructural
localization of polyclonal GABA and monoclonal glutamate
immunoreactivity in the rat midbrain periaqueductal gray. Soc.
Neurosci. Abs. 11: 126.

Csillag, A., Stewart, M.G. and E.M. Curtis (1987) GABAergic structures in the
chick telencephalon: GABA immunocytochemistry combined with light
 142

and electron microscope autoradiography, and Golgi impregnation.

Brain Res. 437: 283-297.

De Biasi, S. and A. Rustioni (1988) Glutamate and substance P coexist in primary

afferent terminals in the superficial laminae of the spinal cord. Proc.
Natl. Acad. Sci. 85: 7820–7824.

Defeudis, F.V. (1983) Central GABAergic systems and analgesia. Drug Dev. Res.
3: 1-15.

Depaulis, A., Morgan, M.M. and J.C. Liebeskind (1987) GABAergic modulation of
the analgesic effects of morphine microinjected in the periaqueductal
gray matter of the rat. Brain Res. 436; 223-228.
Dostrovsky, J.O. and J.F.W. Deakin (1977) Periaqueductal gray lesions reduce
morphine analgesia in the rat. Neurosci. Lett. 4: 99-103.
Fardin, V., Oliveras, J.L. and J.M. Besson (1984) Projections from the
periaqueductal gray matter to the B3 cellular area (nucleus raphe
magnus and nucleus reticularis paragigantocellularis) as revealed by the
retrograde transport of horseradish peroxidase in the rat. J. Comp. Neurol.
223: 483-500.

Gallager, D.W. and A. Pert (1978) Afferents to brain stem nuclei (brain stem
raphe, nucleus reticularis pontis caudal is and nucleus
gigantocellularis) in the rat as demonstrated by microiontophoretically
applied horseradish peroxidase. Brain Res. 144: 257-275.
Hammond, D.L. and E.J. Drower (1984) Effects of intrathecally administered
THIP, baclofen and muscimol on nociceptive threshold. Eur. J.
Pharmacol. 103: 121-125.

Haefely, W. and P. Polc (1986) Physiology of GABA enhancement by

benzodiazepines and barbiturates. In: Benzodiazepine/GABA Receptors
and Chloride Channels: Structural and Functional Properties. Liss. pp.
97-133.

Ho, I.K., Loh, H.H. and E.L. Way (1976) Pharmacological manipulation of
gamma-aminobutyric acid (GABA) in morphine analgesia, tolerance and
physical dependence. Life Sci. 18: 1111-1124.
Hynes, M.D., Leander, J.D., Frederickson, R.C.A., Ho, P.P.K., Johnson, D.W. and
R.A. Archer (1984) Evaluation of THIP in standard tests for analgesic
activity: occurrence of antinociception and hyperalgesia. Drug. Dev. Res.
4: 405-419.

Iversen, L.L. and F.E. Bloom (1972) Studies of the uptake of 3H-GABA and 3H
glycine in slices and homogenates of rat brain and spinal cord by
electron microscope autoradiography. Brain Res. 41: 131-143.
Iversen, L.L. and J.S. Kelly (1975) Uptake and metabolism of Y-aminobutyric
acid by neurones and glial cells. Biochem. Pharmacol. 24: 933-938.
 14 3

Izumi, K., Munekata, E., Yamamoto, H., Nakanishi, T. and A. Barbeau (1980)
Effects of taurine and Y-aminobutyric acid on akinesia and analgesia
induced by D-Ala”-met-enkephalinamide in rats. Peptides 1: 139-146.
Kelly, D.D. and M. Glusman (1968) Aversive thresholds following midbrain
lesions. J. Comp. Physiol. Psychol. 66: 25-34.
Kisvárday, Z.F., Cowey, A., Hodgson, A.J. and P. Somogyi (1986) The relationship
between GABA immunoreactivity and labelling by local uptake of
[3H]GABA in the striate cortex of monkey. Exp. Brain Res. 62: 89-98.
Lam, D.M.K., Su, Y.Y.T. and C.B. Watt (1986) The self-regulatory synapse: a
functional role for the co-existence of neuroactive substances. Brain Res.
Rev. 11: 249-257.

Lauder, J.M., Han, V.K.M., Henderson, P., Verdoon, T. and A.C. Towle (1986)
Prenatal ontogeny of the GABAergic system in the rat brain: an
immunocytochemical study. Neurosci. 19: 465-493.
Lim, C.R., Garant, D.S. and K. Gale (1985) GABA agonist induced analgesia from
the lateral preoptic area in the rat. Eur. J. Pharmacol. 107: 91-94.
Liu, R.P.C. and B.L. Hamilton (1980) Neurons of the periaqueductal gray matter
as revealed by Golgi study. J. Comp. Neurol. 189: 403-418.
Lovick, T.A. (1985) Ventrolateral medullary lesions block the antinociceptive
and cardiovascular responses elicited by stimulating the dorsal
periaqueductal grey matter in rats. Pain 21: 241-252.
Mayer, D.J. and J.C. Liebeskind (1974) Pain reduction by focal electrical
stimulation of the brain: an anatomical and behavioral analysis. Brain
ReS. 68: 73-93.

Mantegazza, P., Tammiso, R., Vicentini, L., Zambotti, F. and N. Zonta (1979)
Muscimol antagonism of morphine analgesia in rats. Br. J. Pharmacol.
67: 103-107.

Mantyh, P.W. (1983) Connections of the midbrain periaqueductal gray in the

monkey. II. Descending efferent projections. J. Neurophysiol. 49: 582
594.

Melzack, R., Stotler, W.A. and W.K. Livingston (1958) Effects of discrete
brainstem lesions in cats on perception of noxious stimulation. J.
Neurophysiol. 21: 353–367.
Moreau, J.L. and H.L. Fields (1986) Evidence for GABA involvement in midbrain
control of medullary neurons that modulate nociceptive transmission.
Brain Res. 397: 37–46.

Morgan, M.M. Levin, E.D. and J.C. Liebeskind (1987) Characterization of the
analgesic effects of the benzodiazepine antagonist, Ro 15-1788. Brain Res.
415: 367-370.
 14.4

Mugnaini, E. and W.H. Oertel (1985) An atlas of the distribution of GABAergic

neurons and terminals in the rat CNS as revealed by GAD
immunohistochemistry, In A. Björklund and T. Hökfelt, (eds.), Handbook
of Chemical Neuroanatomy, Vol. 4, GABA and Neuropeptides in the CNS.
Part I, Elsevier, Amsterdam, pp. 436-608.
Nagai, T., McGeer, T.L., and E.G. McGeer (1983) Distribution of GABA-T intensive
neurons in the rat forebrain and midbrain. J. Comp. Neurol. 218: 220-238.
Nicoll, R.A., Alger, B.E. and C.E. Jahr (1980) Enkephalin blocks inhibitory
pathways in the vertebrate CNS. Nature 287: 22-25.
Nisticó, G., DiGiorgio, R.M., Rotiroti, D., Naccari, F. and A. Calatroni (1979)
Effects of intraventricular 3-endorphin on GABA system in some areas of
chick brain. Res. Comm. Chem. Pathol. Pharmacol. 26: 469-478.

Ottersen, O.P. and J. Storm-Mathisen (1984) lutamate- and GABA-containing

neurons in the mouse and rat brain, as demonstrated with a new
immunocytochemical technique. J. Comp. Neurol. 229: 374-392.
Paxinos, G. and C. Watson (1986) The rat brain in stereotaxic coordinates, 2nd
ed., Academic Press, Orlando, FL.

Prieto, G.J., Cannon, J.T. and J.C. Liebeskind (1983) N. raphe magnus lesions
disrupt stimulation-produced analgesia from ventral but not dorsal
midbrain areas in the rat. Brain Res. 261 : 53-57.

Proudfit, H.K. and R.A. Levy (1978) Delimitation of neuronal substrates

necessary for the analgesic action of baclofen and morphine. Eur. J.
Pharmacol. 47: 159-166.

Reichling, D.B. and A.I. Basbaum (1987) Anatomical evidence that GABA
influences the projection from the periaqueductal gray matter to the
nucleus raphe magnus. Pain, Supp. 4: S40. -

Reichling, D.B., Cho, H.J. and A.I. Basbaum (1988) GABA-immunoreactive

synaptic contacts onto projection neurons of the periaqueductal gray
matter and the nucleus raphe magnus. Soc. Neurosci. Abs. 14: 856.
Reichling, D.B., Kawashima, Y. and Basbaum, A.I. (submitted) GABA, 3
endorphin, enkephalin, and dynorphin immunore activity in the
periaqueductal gray of the rat. Synapse.
Retz, K.C. and L.M. Holaday (1986) Analgesia and motor activity following
administration of THIP into the periaqueductal gray and lateral ventricle.
Drug. Dev. Res. 9; 133-142.

Rhodes, D.L. (1979) Periventricular system lesions and stimulation-produced

analgesia. Pain 7: 51-63.
Romandini, S. and R. Samanin (1984) Muscimol injections in the nucleus raphe
dorsalis block the antinociceptive effects of morphine in rats: apparent
 1 4.5

lack of 5-hydroxytryptamine involvement in muscimol's effect. Br. J.

Pharmacol. 81: 25-29.

Sandkühkler, J. and G.F. Gebhart (1984) Relative contributions of the nucleus

raphe magnus and adjacent medullary reticular formation to the
inhibition by stimulation in the periaqueductal gray of a spinal
nociceptive reflex in the pentobarbitol-anesthetized rat. Brain Res. 305:
77-87.

Sandner, G., Dessort, D., Schmitt, P. and P. Karli (1981) Distribution of GABA in
the periaqueductal gray matter. Effects of medial hypothalamic lesions.
Brain Res. 224: 279-290.

Sandner, G., Schmitt, P. and P. Karli (1986) Unit activity alterations induced in
the mesencephalic periaqueductal gray by local electrical stimulation.
Brain Res. 386: 53-63.

Sawynok, J. (1983) Monoamines as mediators of the antinociceptive effect of

baclofen. Naunyn-Schmiedeberg's Arch. Pharmacol. 323: 54-57.

Schon, F. and L.L. Iversen (1972) Selective accumulation of [3H]GABA by

stellate cells in rat cerebellar cortex in vivo. Brain Res. 42: 503-507.

Sternberger, L.A. (1979) Immunocytochemistry, 2nd ed., John Wiley and Sons,
New York.

Tredici, G., Bianchi, R. and M. Gioia (1983) Short intrinsic circuit in the
periaqueductal gray matter of the cat. Neurosci. Lett. 39: 131-136.
Ueda, H., Ge, M., Satoh, M. and H. Takagi (1987) Subconvulsive doses of
intracisternal bicuculine methiodide, a GABAA receptor antagonist,
produce potent analgesia as measured in the tail pinch test in mice. Euro.
J. Pharmacol. 136: 129-131.

Vaught, J.L., Pelley, K., Costa, L.G., Setler, P. and S.J. Enna (1985) A comparison
of the antinociceptive responses to the GABA-receptor agonists THIP and
baclofen. Neuropharmacol. 24: 211-216.
Watt, C.B., Li, T., Lam, D.M.K. and S.M. Wu (1988) Quantitative studies of
enkephalin's coexistence with g-aminobutyric acid, glycine and
neurotensin in amacrine cells of the chicken retina. Brain Res. 444; 366
370.

Wilson, P.R. and T.L. Yaksh (1978) Baclofen is antinociceptive in the spinal
intrathecal space of animals. Euro. J. Pharmacol. 51: 323-330.
Yaksh, T.L. and T.A. Rudy (1978) Narcotic analgetics: CNS sites and mechanisms
of action as revealed by intracerebral injection techniques. Pain 4: 299
359.

Yaksh, T.L., Yeung, J.C. and T.A. Rudy (1976) Systematic examination in the rat
of brainstem sites sensitive to the direct application of morphine:
 146

observation of differential effects within the periaqueductal gray. Brain

ReS. 114: 83-103.

Yeung, J.C., Yaksh, T.L. and T.A. Rudy (1975) Effects of brain lesions on the
antinociceptive properties of morphine in rats. Clin. Exp. Pharmacol.
Physiol. 2: 261-268.
Young, W.S. and M.J. Kuhar (1980) Radiohistochemical localization of
benzodiazepine receptors in rat brain. J. Pharmacol. Exp. Thera. 212:337
346.

Zambotti, F., Zonta, N., Parenti, M., Tommasi, R., Vicentini, L., Conci, F. and P.
Montegazza (1982) Periaqueductal gray matter involvement in the
muscimol induced decrease of morphine antinociception. Naunyn
Schmiedeberg's Arch. Pharmacol. 318; 368-369.
Zieglgänsberger, W., French, E.D., Siggins, G.R. and F.E. Bloom (1979) Opioid
peptides may excite hippocampal pyramidal neurons by inhibiting
adjacent inhibitory interneurons. Science 205: 415.
Zonta, N., Zambotti, F., Vicentini, R., Tammiso, R. and P. Mantegazza (1981)
Effects of some GABA-mimetic drugs on the antinociceptive activity of
morphine and b-endorphin in rats. Naunyn-Schmiedeberg's Arch.
Pharmacol. 316: 231-234.

Zorn, S.H. and S.J. Enna (1987) The GABA agonist THIP, attenuates
antinociception in the mouse by modifying central cholinergic
transmission. Neuropharmacol. 26: 433-437.

Zucker, C., Yazulla, S. and J.-Z. Wu (1984) Non-correspondence of [3H] GABA

uptake and GAD localization in goldfish amacrine cells. Brain Res. 298:
154-158.
 14.9

Fig. 2. This dark-field photomicrograph shows the distribution of retrogradely

labeled neurons in the PAG after an injection of WGAapo HRP-Au in the NRM.
Labeled neurons (small, white spots) are common in most of the PAG, but are
rare in the dorsolateral sector (asterisk) and in the dorsal raphe nucleus
(DRN). Scale bar = 0.5 mm.
 1 5 1

Fig. 3. This photomicrograph shows the light-microscopical appearrance of

retrograde tracer and pre-embedding GABA immunore activity in the
ventrolateral PAG, before osmication and embedding. Retrogradely labeled
neurons (arrows) contain numerous black particles of silver-enhanced tracer
in their cell body and proximal dendrites. GABA-immunoreactive terminals
(arrowheads) appear as small, brownish punctae in the DAB-reacted tissue.
Scale bar = 100 plm.
 º º - -------
- --
*º-º-º-º-
-- --- ----
-. :
- -- - -

*
º
-- -

--- --- - -- ----

--- º Sºº- *- º º: sº
-- -

º
- -
-
- - -

---
º -º-º-º: wº º º: ******
--
-
--- -

-
--
-

- ** ºº
------." º -
--
>

tº Sº
- -

- - -

º
- - - - - --- - -- -
tºº-º-º-º-º-
- - - - -
-
** * *

ºº: ºº - ºº:
ºº: ºrº - - - - - * ,

- * : *
*** º
º --------->
-
º: --- --- -
---
--

"----, -
-
-
-
-
-

+-sº
- -

-
--
 153

Fig. 4. This figure shows the electron microscopic appearance of post

embedding GABA immunoreactivity in the ventrolateral PAG. GABA
immunore activee terminals (g 1 and g2) contain numerous particles of
colloidal gold (arrowheads). The GABA-immunoreactive terminal, g1, contains
dense-cored vesicles (open arrows) in addition to small, clear, pleomorphic
vesicles. This terminal forms a symmetrical synaptic contact (arrow) with an
unlabeled dendrite (den). An adjacent unlabeled axon terminal (a) contains
few particles of colloidal gold. Scale bar = 0.5 pum.
 155

Fig. 5. This micrograph shows a thin dendrite (den) that receives particularly
dense GABA-immunoreactive synaptic inputs. In this plate, five GABA
immunore active terminals (g) make symmetrical synaptic contacts
(arrowheads) onto the dendrite. Other, nearby axon terminals (a) are
unlabeled. Scale bar = 0.5 pum.
 157

Fig. 6. Axon terminals in the ventrolateral PAG are often separated from other
neuronal processes by glial processes. In this figure, a GABA-immunoreactive
(g 1) and an unlabeled terminal (a) are separated from a cell body (cb) by a
very thin astrocytic sheath (arrowheads). At the light-microscopic level,
these terminals would appear to be directly apposed to the cell body. Also in
this figure are another GABA-immunoreactive axon terminal (g2) and a GABA
immunoreactive unmyelinated axon or preterminal segment (arrow). Scale
bar = 0.5 pum.
■±√■
·
 1 5 9

Fig. 7. These electron micrographs show tissue stained by pre-embedding

immunocytochemistry. Neuronal cell bodies (cb) that contain silver
enhanced and gold-toned particles of WGAapo HRP-Au (arrows). This gold
conjugated tracer has been retrogradely transported from the NRM. The
labeled cell bodies are synaptically contacted by GABA-immunoreactive axon
terminals (arrowheads). These terminals are immunostained by a flocculent,
electron-dense DAB reaction product. Scale bars = 1 pum and 0.5 pum in A and B,
respectively.
 1 6 1

Fig. 8. These electron micrographs illustrate GABA-immunoreactive axon

terminals (g), stained by a pre-embedding procedure, that are presynaptic to
retrogradely labeled cell bodies in the ventrolateral PAG. The latter are
identified by the presence of silver-enhanced and gold toned WGAapo HRP-Au
particles (arrows) within lysosomes. (The tracer was injected in the NRM.)
Arrowheads point to synaptic specializations between the labeled terminals
and cell bodies. Scale bars = 0.5 pum.
 163

Fig. 9. This figure illustrates a neuronal cell body (CB) in the ventrolateral
PAG which is post-synaptic to unlabeled axon terminals (arrowheads) and
GABA-immunoreactive axon terminals (arrows). The latter were stained by
post-embedding immunocytochemistry. The cell body contains silver
enhanced and gold-toned particles of WGAapo HRP-Au retrogradely tansported
from the NRM. The nucleus is labeled "Nu." The two regions outlined by
squares are shown at higher magnification in figure 10. Scale bar = 1 p.m.
 1.65

Fig. 10. This figure shows the regions outlined in figure 9 at higher
magnification. GABA-immunoreactive axon terminals (g), stained by the post
embedding procedure, make synaptic contacts (arrowheads) onto the cell body
(CB). Plate B also shows an unlabeled axon terminal (a) which is presynaptic
to the cell body. Scale bars = 0.5 p.m.
 167

CHAPTER VI

Electron Microscopic Immunocytochemical Evidence for Synaptic

Interactions
between GABAergic Terminals and Serotonergic Neurons
in the Dorsal Raphe Nucleus of the Rat

David B. Reichling, Geneviève Chazal, and Allan I. Basbaum

 168

Pharmacological and physiological evidence suggests that serotonergic neurons of

the dorsal raphe nucleus (DRN) are under GABAergic control. For example, ionophoretic
application of Y-aminobutyric acid (GABA) in the DRN inhibits units with firing patterns
characteristic of serotonergic neurons [41]. Furthermore, the GABA antagonist,
picrotoxin, blocks the inhibition of DRN serotonergic neurons that is produced by systemic
administration of antipsychotic drugs or by electrical stimulation of the habenula or pontine
reticular formation [11, 12, 40, 41]. Since presumptive GABAergic axon terminals in the
DRN have been detected by GABA [7] or glutamic acid decarboxylase (GAD)
immunocytochemistry [23], or by uptake of [3H]GABA [4], the picrotoxin-sensitive
suppression of activity in the DRN might be mediated by GABAergic terminals that
synapse directly onto 5-HT-containing neurons. In fact, Harandi et al. [15] noted that
some terminals that accumulate [3H]GABA are in close apposition to, or synapse with, 5
HT-immunoreactive dendrites. The fact that a neuron accumulates extracellular GABA,
however, does not assure that it synthesizes GABA as a transmitter -- neurons that are not
GABA-immunoreactive can accumulate [3H]GABA [8,18,43], and even glia possess high
affinity GABA uptake mechanisms [17]. To avoid this problem, in the present study we
used simultaneous electron microscopic immunocytochemical detection of GABA and 5
HT to provide direct anatomical evidence of synaptic contacts between presumptive
GABAergic and serotonergic neuronal elements in the DRN. GABA-immunoreactive
profiles were immunostained by the peroxidase-anti-peroxidase method [38]; 5-HT
immunoreactive profiles were immunolabeled with colloidal gold [19].
Male Sprague-Dawley rats were perfused with 100ml of 0.05M phosphate-buffered
saline (PBS) followed by 500ml of fixative containing 4% paraformaldehyde and 1%
glutaraldehyde in PBS. The brains were removed and 30 pum sections were cut on a
Vibratome. The sections were first incubated, for one hour, in a blocking solution of 3%
normal goat serum in PBS, and then for 24 hours in a rabbit antiserum raised against a
BSA conjugate of 5-HT (Incstar Corp.). The sections were next incubated in goat-anti
 169

rabbit IgG labeled with 15nm colloidal gold particles (Janssen). After extensive washing,
sections were incubated in a polyclonal rat anti-GABA antiserum, diluted 1:3750. This
antiserum (kindly provided by Dr. Andrew Towle of Cornell University) is directed against
a glutaraldehyde conjugate of GABA and bovine serum albumin. Radioimmunoassay
shows that the antiserum is highly specific for glutaraldehyde-fixed GABA [20].
Prolonged incubation times were used to improve penetration of antibody and avoid the
ultrastructural damage caused by detergents. Thus, after 48 hours, the sections were
incubated in goat-anti-rat IgG (1:100; Cappel) for 12 hours, followed by 1 hour in rat
peroxidase-anti-peroxidase (1:500; ICN Immunobiologicals). GABA immunoreactivity
was visualized by reacting the tissue with diaminobenzidine (DAB). Some of these sections
were silver-enhanced and gold-toned to improve visibility of the colloidal gold particles [2].
The tissue was osmicated, dehydrated, and embedded in Polybed 812 (Polysciences).
Ultrathin sections were cut from the surface of the tissue and lightly counterstained with
lead citrate and uranyl acetate. Preincubation of either antiserum with 10 puM of the
appropriate immunogen completely blocked immunostaining in the DRN.
At the light microscopic level, 5-HT immunoreactivity appears as a light, pinkish
Stain over cell bodies and dendrites in the DRN. GABA-immunoreactive terminals are

distributed throughout the DRN, and are densest in the dorsal part of the nucleus. At the
electron microscopic level, 5-HT-immunoreactive terminals contain scattered clusters of
electron-dense colloidal gold particles (Figure 1). GABA immunoreactivity consists of a
dark, diffuse and flocculent stain, within vesicle-containing profiles and axons. Gold- and
DAB-labeled profiles were considered positively labeled only if immunostaining was
evident in serial sections. We detected very few GABA-immunoreactive cell bodies or
dendrites, probably because the concentration of transmitter in these structures is below the
detectable limit of our methods.

Most GABA-immunoreactive terminals were associated with unlabeled dendrites,

but some contacted 5-HT immunoreactive dendritic profiles (Figure 1a). The presynaptic
 170

GABA-immunoreactive terminals contained densely-packed, small, clear vesicles. Pre

and post-synaptic specializations were often present; when the presynaptic densities were
not obscured by dense DAB reaction product, the synaptic morphology appeared to be
symmetrical, Gray Type II [14]. In many cases, dendrites contacted by GABA
immunoreactive terminals were also contacted by other, unlabeled, terminals (Figure 1b);
this indicates that chemically diverse inputs converge onto single dendrites of 5-HT
Il CuTOITS.

Previous studies reported that 5-HT and GABA coexist in some cell bodies,
dendrites, and terminals in the DRN [5, 13, 24]. Those double-labeling studies used
immunocytochemical localization of glutamate decarboxylase, GABA, or 5-HT, combined
with autoradiographic detection of [3H]GABA or [3H]5-HT accumulation. In our studies,
the vast majority of terminals were only single-labeled. However, in a few cases we found
axons and terminals immunoreactive for both 5-HT and GABA (Figure 2). On the other
hand, although dense-cored vesicles are found in 50% of GABA-immunoreactive terminals
in the ventral PAG [30], we never observed dense-cored vesicles in GABA
immunoreactive terminals that contacted 5-HT-immunoreactive dendrites. It is, therefore,
unlikely that GABA coexists with peptide transmitters at these synapses.
Although GABA cell bodies were not labeled in the present study, presumptive
GABAergic neurons have been detected in the DRN [4, 23, 29, 31]. It is thus likely that at
least some of the GABA-immunoreactive terminals that we observed originate from local
interneurons. Consistent with this idea, lesions of major sources of afferents to the DRN
(the dorsal pontine tegmentum, substantia nigra, or habenula) did not alter GAD activity in
the DRN [4]. The fact that only a few GABAergic neurons contribute to the major efferent
projection from the PAG to the ventromedial medulla [31] is also consistent with their
being interneurons.
GABA-immunoreactive inputs to 5-HT-immunoreactive neurons probably
contribute to the controls that influence ascending serotonergic projections from the DRN.
 171

For example, systemic enhancement of GABAergic transmission reduces 5-HT synthesis

and turnover in areas that receive 5-HT projections from the DRN, including striatum,
frontal cortex, hypothalamus, and substantia nigra [25, 26, 35]. Furthermore, since
infusion of GABA, or the GABAA receptor agonist, muscimol, directly into the DRN
reduces 5-HT metabolism in the striatum [28, 36], the effect of systemic GABA upon
striatal 5-HT probably occurs, in part, within the DRN.
The anxiolytic action of benzodiazepines is thought to result, in part, from thier
enhancement of GABAergic inhibition of 5-HT neurons in the dorsal and median raphe
nuclei [37]. Benzodiazepines interact with the GABAA receptor complex to enhance the
inhibitory effect of GABA (c.f. 16). Benzodiazepine agonists applied iontophoretically in
the DRN potentiate the inhibition of 5-HT neurons caused by intophoretically applied
GABA [10]. Furthermore, the observation that benzodiazepine agonists alone inhibit
spontaneous activity of DRN 5-HT neurons implies a tonically active GABAergic input
[21, 39]. On the other hand, the fact that picrotoxin does not alter the firing rate of these
neurons, implies that the endogenous GABAergic inputs are not tonically active [10].
The GABA/5-HT interaction in the DRN might also be related to cardiovascular
regulation. Electrical stimulation of the DRN elevates blood pressure in rats. Since
pretreatment with 5,7-dihydroxytryptamine reduces this pressor effect, 5-HT neurons are
probably involved [32]. Microinjection of muscimol, but not baclofen (a GABAB
agonist), into the DRN lowers heart rate and blood pressure [33]. Finally, since
microinjected picrotoxin does not cause cardiovascular changes, the relevant GABAergic
synapse in the DRN is probably not tonically active [33].
Serotonergic neurons of the DRN have also been implicated in antinociceptive
mechanisms. Some serotonergic DRN neurons project to the medullary nucleus raphe
magnus [3] from which originates a well-characterized descending inhibitory projection
[c.f. 1]. The DRN may be the most efficacious region in the PAG for eliciting electrical
stimulation-produced analgesia [27], and DRN lesions antagonize the analgesic effect of
 172

systemic morphine [34, 42]. Since picrotoxin microinjected in this region produces
analgesia, and muscimol reduces the analgesia produced by morphine [22], it is possible
that this projection is subject to GABAergic controls. The ligands that produced effects in
all the studies discussed in the paragraphs above, were specific for the GABAA receptor.
Thus, it seems likely that the effect of GABA upon serotonergic neurons in the DRN, are
mediated by GABAA receptors. Consistent with this idea, GABAA binding sites are dense
in the DRN, but GABAB receptors are sparse [6].
In conclusion, the present study demonstrates that 5-HT-immunoreactive neurons
in the DRN are contacted by GABA-immunoreactive terminals. There is evidence that this
GABAergic control of serotonergic projections from the DRN is involved in the actions of
benzodiazepines, cardiovascular regulation, and antinociception. In addition, since the
widely divergent serotonergic projections from the DRN [9] have been implicated in a
variety of other functions including hormone regulation, locomotion, sleep, eating, sexual
activity, and aggression, it seems likely that many of these functions are also subject to
GABAergic controls acting upon 5-HT neurons in the DRN.

ACKNOWLEDGEMENTS

The authors thank Ms. Bonnie Lord for technical assistance, and Ms. Simona Ikeda for
photographic assistance. This work was supported by NIH Public Health Service Grants
NS 14627 and 21445.
 17 3

REFERENCES

1. Basbaum, A.I. and Fields, H.L., Endogenous pain control systems: brainstem spinal
pathways and endorphin circuitry, Ann. Rev. Neurosci., 7 (1984) 309-338.
. Basbaum, A.I. and Menetrey, D., Wheat germ agglutinin-apoh RP gold: a new
retrograde tracer for light- and electron-microscopic single- and double-label
studies, J. Comp. Neurol., 261 (1987) 306–318.
. Beitz, A.J., Shepard, R.D. and W.E. Wells (1983) The periaqueductal gray-raphe
magnus projection contains somatostatin, neurotensin and serotonin but not
cholecystokinin. Brain Res. 261: 132-137.
. Belin, M.F., Aguera, M., Tapaz, A., McRae-DeQueurce, A., Bobillier, P. and Pujol,
J.F., GABA-accumulating neurons in the nucleus raphe dorsalis and periaqueductal
gray in the rat: a biochemical and radioautographic study, Brain Res., 170 (1979)
279-297.

. Belin, M.F., Nanopoulos, D., Didier, M., Aguera, M., Steinbusch, H., Verhofstad,
A., Maitre, M. and Pujol, M.F., Immunohistochemical evidence for the presence of
Y-aminobutyric acid and serotonin in one nerve cell. A study on the raphe nuclei of
the rat using antibodies to glutamate decarboxylase and serotonin, Brain Res., 275
(1983) 329-339.
. Bowery, N.G., Hudson, A.L. and Price, G.W., GABAA and GABAB receptor site
distribution in the rat central nervous system, Neurosci. 20 (1987) 365-383.
. Clements, J.R., Beitz, A.J., Larson, A.A. and Madl, J.E., The ultrastructural
localization of polyclonal GABA and monoclonal glutamate immunoreactivity in the
rat midbrain periaqueductal gray, Soc. Neurosci. Abs., 11 (1985) 126.
. Csillag, A., Stewart, M.G. and E.M. Curtis (1987) GABAergic structures in the chick
telencephalon: GABA immunocytochemistry combined with light and electron
microscope autoradiography, and Golgi impregnation. Brain Res. 437: 283-297.
. Dahlström, A. and Fuxe, K., Evidence for the existence of monoamine-containing
neurons in the central nervous system. I. Demonstration of monoamines in the cell
bodies of brainstem neurons, Acta Physiol. Scand., 62 (1964) 1-55.
10. Gallager, D.W., Benzodiazepines: potentiation of a GABA inhibitory response in the
dorsal raphe nucleus, Euro. J. Pharmacol. 49 (1978) 133-143.
11. Gallager, D.W. and Aghajanian, G.K., Effect of antipsychotic drugs on the firing of
dorsal raphe cells. I. Role of adrenergic system, Eur. J. Pharmacol., 39 (1976)
341-355.

12. Gallager, D.W. and Aghajanian, G.K., Effect of antipsychotic drugs on firing rate of
dorsal raphe cells. II. Reversal by picrotoxin, Eur. J. Pharmacol., 39 (1976) 357
364.
 1 74

13. Gamrani, H., Harandi, M., Belin, M.F., Dubois, M.P., and Calas, A., Direct
electronmicroscopic evidence for the coexistence of GABA uptake and endogenous
serotonin in the same rat central neurons (by coupled radioautographic and
immunocytochemical procedures), Neurosci. Lett., 48 (1984) 25-30.
14. Gray, E.G., Axo-somatic and axo-dendritic synapses of the cerebral cortex: An
electronmicroscope study, J. Anat. (1959) 93: 420-433.
15. Harandi, M., Aguera, M., Gamrani, H., Didier, M., Maitre, M., Calas, A. and Belin,
M.F., Y-aminobutyric acid and 5-hydroxytryptamine interrelationship in the rat
nucleus raphe dorsalis: combination of radioautographic and immunocytochemical
techniques at light and electron microscopy levels, Neurosci., 21 (1987) 237-251.
16. Haefely, W. and Polc, P., Physiology of GABA enhancement by benzodiazepines and
barbiturates, in: Benzodiazepine/GABA receptors and chloride channels: Structural
and Functional Properties, Liss 1986, pp.97-133.
17. Iversen, L.L. and Kelly, J.S., Uptake and metabolism of Y-aminobutyric acid by
neurones and glial cells, Biochem. Pharmacol., 24 (1975) 933-938.
18. Kisvárday, Z.F., Cowey, A., Hodgson, A.J. and Somogyi, P., The relationship
between GABA immunoreactivity and labelling by local uptake of [3H]GABA in the
striate cortex of monkey, Exp. Brain Res.62 (1986) 89–98.
19. Lamberts, R. and Goldsmith, P.C., Preembedding colloidal gold immunostaining of
hypothalamic neurons: light and electron microscopic localization of 3-endorphin
immunoreactive perikarya, J. Histochem. Cytochem., 33 (1985) 499-507.
20. Lauder, J.M., Han, V.K.M., Henderson, P., Verdoorn, T. and Towle, A.C., Prenatal
ontogeny of the GABAergic system in the rat brain: an immunocytochemical study,
Neurosci., 19 (1986) 465-493.
21. Laurent, J.P., Mangold, M., Humbel, U. and Haefely, W., Reduction by two
benzodiazepines and pentobarbitone of the multiunit activity in substantia nigra,
hippocampus, nucleus locus coeruleus and nucleus raphé dorsalis of encéphale
isolé rats, Neuropharmacol. 22 (1983) 501-511.
22. Moreau, J.-L. and Fields, H.L., Evidence for GABA involvement in midbrain control
of medullary neurons that modulate nociceptive transmission, Brain Res., 397 (1986)
37–46.

23. Mugnaini, E. and Oertel, W.H., An atlas of the distribution of GABAergic neurons
and terminals in the rat CNS as revealed by GAD immunohistochemistry, in A.
Björklund and T. Hökfelt (eds.), Handbook of Chemical Neuroanatomy, Vol. 4,
GABA and Neuropeptides in the CNS, Part I, Elsevier, Amsterdam, 1985, pp. 436
608.

24. Nanopoulos, D., Belin, M.F., Maitre, M., Vincendon, G. and Pujol, J.F.,
Immunocytochemical evidence for the existence of GABA-ergic neurons in the
nucleus raphe dorsalis. Possible existence of neurons containing serotonin and
GABA, Brain Res., 232 (1982) 375-389.
 175

25 . Nishikawa, T. and Scatton, B., Evidence for a GABAergic inhibitory influence on

serotonergic neurons originating from the dorsal raphe, Brain Res., 279 (1983)
325-329.

26 . Nishikawa, T. and Scatton, B., Inhibitory influence of GABA on central

serotoninergic transmission raphe nuclei as the neuroanatomical site of the
GABaergic inhibition of cerebral serotoninergic neurons, Brain Res., 331 (1985)
91-103.

27. Oliveras, J.L., Besson, J.M. and G. Guilbaud (1974) Behavioral and
electrophysiological evidence of pain inhibition from midbrain stimulation in the
cat. Exp. Brain Res. 20:32-44.
28. Przewlocka, B., Stala, L. and Scheel-Krüger, J., Evidence that GABA in the nucleus
dorsalis raphe induces stimulation of locomotor activity and eating behavior, Life
Sci., 25 (1979) 937–946.
29. Reichling, D.B. and Basbaum, A.I., Anatomical evidence that GABA influences the
projection from the periaqueductal gray matter to the nucleus raphe magnus, Pain,
Supp. 4 (1987) S40.
30. Reichling, D.B., Cho, H.J. and Basbaum, A.I., GABA-immunoreactive synaptic
contacts onto projection neurons of the periaqueductal gray matter and the nucleus
raphe magnus, Soc. Neurosci. Abs., Vol. 14, Part 2 (1988) 856.
31. Reichling, D.B., Kawashima, Y. and Basbaum, A., GABA, beta endorphin,
enkephalin, and dynorphin immunoreactivity in the rat midbrain periaqueductal
gray, submitted for publication.
32. Robinson, S.E., Austin, M.J.F. and Gibbens, D.M., The role of serotonergic
neurons in dorsal raphe, median raphe and anterior hypothalamic pressor
mechanisms, Neuropharmacol. 24 (1985) 51-58.
33. Robinson, S.E., Rice, M.A. and Davidson, W., A GABA cardiovascular mechanism
in the dorsal raphe of the rat, Neuropharmacol. 25 (1986) 611-61.5.
34. Samanin, R., Gumulka, W. and L. Valzelli (1970) Reduced effect of morphine in
midbrain raphe lesioned rats. Eur. J. Pharmacol. 10: 339-343.
35. Scatton, B., Zivkovic, B., Dedek, J., Lloyd, K.G., Constantinidis, J., Tissot, R. and
Bartholini, G., Y-aminobutyric acid (GABA) receptor stimulation. III. Effect of
progabide (SL 76002) on norepinepherine, dopamine and 5-hydroxytryptamine
turnover in rat brain areas, J. Pharmacol. Exp. Ther., 220 (1982) 678-688.
36. Scatton, B., Serrano, A., Rivot, J.P. and Nishikawa, T., GABAergic inhibitory
influence on striatal serotoninergic transmission exerted in the dorsal raphe as
revealed by in vivo voltametry, Brain Res., 305 (1984) 343-352.
37. Stein, L., Wise, C.D. and Beluzzi, J.D., Effects of benzodiazepines on central
serotonergic mechanisms, Adv. Biochem. Psychopharmacol., 14 (1975) 29-44.
38. Sternberger, L.A., Immunocytochemistry, 2nd ed., New York, John Wiley and
Sons, 1979.
 176

39. Trulson, M.E., Preussler, D.W., Howell, G.A. and Frederickson, C.J., Raphe unit
activity in freely moving cats: effects of benzodiazepines, Neuropharmacol. 21
(1982) 1045-1050.
40. Wang, R.Y., Gallager, D.W. and Aghajanian, G.K., Stimulation of pontine reticular
formation suppresses firing of serotonergic neurones in the dorsal raphe, Nature,
264 (1976) 365-368.
41. Wang, R.Y. and Aghajanian, G.K., Physiological evidence for habenula as major link
between forebrain and midbrain raphe, Science, 197 (1977) 89-91.
42. Yaksh, T.L., Plant, R.L. and Rudy, T.A. (1977) Studies on the antagonism by raphe
lesions of the antinociceptive action of systemic morphine. Eur. J. Pharmacol. 41:
399-408.

43. Zucker, C., Yazulla, S. and Wu, J.Y., Non-correspondence of [3H]GABA uptake and
GAD localization in goldfish amacrine cells, Brain Res., 298 (1984) 154-158.
 17 7

Fig. 1. A,B: Electronmicrographs of GABA-immunoreactive terminals (asterisks) in

synaptic contact with 5-HT-immunoreactive dendrites (Den). The labeled dendrites contain
scattered colloidal gold particles (arrowheads). B shows a dendrite contacted by three
GABA-immunoreactive terminals and apposed by an unlabeled terminal (t). Scale bars =
1um.
|× º º
º -
 17 9

Fig. 2: Electronmicrographs of profiles labeled for both GABA and 5-HT. Some small
diameter unmyelinated (A) and myelinated axons (B) are double-labeled. C shows a
double-labeled terminal (asterisk) in synaptic contact with a 5-HT-immunoreactive dendrite
(Den). The dendrite is also contacted by two unlabeled terminals (arrows). Scale bars =
0.5 um.
 º - -
---
---
------
-
---
-
--- -

-- -
→ *, * sº.
~1. - *** * * * * ** * Q- 2.

O/2, sº LièRaº, sº º ji
-

*) C 1. ~ --- * -

L. J
º | || sº ~/C 9. […] º | | °o 3° L. J ‘o.
[…] sº
*~~
- -

º º, sº 9 |T| %2. is ºvº gº & º

/ º ~ :-
-

º, sº *2 sº º
** – º
º ºs
%. s. ■

sº tº
Sºme
"/777/7C■, ■co ºs
S-2 oºglº■■º () ºy- i.Sº (1//, /7.7/
'o.
º,+. ºr nº
º
AT:

—- 2. -

3 *, sº º *

sº º "/777/10, ■co sº ºv
__` 2– ºsº […] º, 9. L. BRARY cºSº [...] º, cº, O/le sº %. LIBRARY º º +- º,
*

[…] “… [...] sº
-

- &
ºgº o, L. J o”
ºs ºc L– º º Q-
&%, […]’ºsº sººvº _\-
gº º,
Q

- %, sº S.
*~ is ºpiº A//yº º
sº º, C■.ºnci■co sººº, cºlº/”0º °sº* C■.7///7.1/■cºw
-

- C -
-

*
~ - - … - - <\
-- *, S
L. J º, J/2 . > sº- ”, Li BRARY sº
*
+, C a. 9.

*.o,L■ B º
--

| | sº
*. º sºvº gº º --
C, -

<2 Aº o, ~

ºvugin * --
C.
ZC ~/C %.!---s
-* ~ sº ** *

[…]
-

~-incºro sº% -ºº º sº,

1, ºr 72, S & c
C■.
-T
-
*2 Sº cºyºcºco $7,
*
*z, * cºpiº*/ºSº %,º i.sº §!//, - ~ - * - - a - -
■

*** R_Y sº L. °. ..)/le s [+º, Li B RA R_Y sº tºº. O/2- sº %. LI

º º 3. *—

%. 'sº Aºvº; G | T * -- s “I■C C, […]

-

~/C
*

º ºº
º º N. **
º

*~~
== *.
C■.º/rºncºco *ºf* *
ºoº- º, sº ºvº gº º,
: º sº cº ºnci■co 3. > cºlº/º a
->
º
-

-- a} *S *, > */
%, L. B RARY S.
e-
- «.
fº / º º Ll &
—- *3 ºx.O
| |
~& Cº- L. J °, ..)[…]le sº […] ”, c B RARY
| | &
º /*

%, O)n -

5 sº
~~

~* ~* 'o
-:/C ~/C º […] cºsº
-r C O
's
*
~
~Tºf 3 IT º,º, %, vº■ 9 IT %,2
~/º & AQ■ KT~:
N. % ~ &sº º

ºs *s onºpia. º %,º º sº
º º/ºrci■co 92 º s
2/ls sº LièRAR's sº ■º
*A* ///7.1//c■,■o sº 4.
O // sºlº * 12 ~
N

sº
Tº sº. […]”. Lºlº sº [] *,
v. &
tºº."º
*c.%, [...] sºsº ºvº G 11 *…°. [...] ■sº
Edº 2 *: Q. **
º

---

sº 17 º º / %.%2 º N. - -

72 -S * S. 'º y

C■.º/7.lºcºco sº*S°, cºol, º/” º %, sº º º/7c, 0 *

Nº. *.

~y. Sºº 0.2%)//11 º

1. - 42 º
-
t
11. (l/ 2. ~.
SS*2 Nº4.
…” * - * - z - *

– ■º,º, C/2-,
--
sº tº tiereº sº. One º ljóº *2
ts 9. *
º

[…] _º *o sº L. ‘o. sº […] *3 *O &

Z/C ”. s Jºjvº 9 IT * Lºdsº& ~/C º%,'— sº<2 ºvºi g 1-1 * % Lºl ºis cº
* -
-

/7-incºco º sººº º/º sº

42, N- 72 º º

~ 7/7/~/ Cº.
42 Sº C■. ºn º/”wº(C '4º,
cºinci■co º,sº,sº 0.05'1.71/11//////■ 2S
& Yºº
2, . --~~%

■º. Jº
- º - -

tº tº 22- - iº — i.
…

S Z- -
• -*

—r -> Q- C■ º ov C

––ls.* -AC º, %, [ ] sºs ºvº■ G 11 * L. J sº S. ºvuginº.*

* º, C■.º/7 ºncºco *s
* S. -
o -

onºpf
-

°o
<>
º ºp1/11///º
/11/ / /
sº cº,
42, sº -
*

Sº
- 2 <
i■/■/■o
z

º, ºut■º
*1, sº
º -
* -

- ■º
1-

--sº.”***
º

[+º, -ºººº
LJB RA R Y
sº. On , .
S
Jo

LC sº
22
J/2 sº sº [] *,
& «. º c Z- A- -

, c
- CT s Jºvº■ %. -- ~/C * I S. * .
o” º, ~ * & O c *o c ~
_-

TIC
*/º ºº,sº sº C■.cºinci■co
2.
Jº■ 9. | TI “, 9 IT º
*,ºsº 0.7%)1/11//?
*/º ”,2. sº 42 S- - ", , ,

sº * Sººn, 1■t J s &

! - _
* - - -- * > (...º.).
º º sº º,
-- *
[…] º, *
RARY § L. º, ..)/le º […] °, Li BRARY Sºº |- º
~
*O
L J B RA
sº
º º
Qe
o O
cº
- I -r-, º --
... * .
:-

* --
~ *
| | | | o” •o
~ --

S º º, º
.C.,* (IC - 7–C -
*/

A..… * *s-- ºvugin º-' s

-
o *
* Qj vºj g : T
* *
ºut!/?wo ºf -
42, N- -
12 º' cºm/ºn- 42 sº -74. -
->
(Y -
º
-
 sº.
§ { °o *~’
C
sº tº ** Sº ”.
L. BRARY
[] º,
º º
& C-
N.S
| |
*J%
º O/2 sº Lié RA º ºsº L. º, Oc O// > sº s
Q-
* */[]sº sº ºvºgri * ”,-- s
* º
-> C

| gº º”,L. sºJ~*
j/)
-/C sº
* (IC * [I] sº s
º
■º sººº, c º/rºncºco *

S
ºut tº sº 25%, cºncº
z
*

|- st-
* . ..
&

C
Q.
tº
L J B R A R_Y
sº
º
Luº. O/2_ º [+º, L. BRARY sº ~, Sº L. J º, Q
[] sºsº ºvº gº °2º, L. J sº * (IC […]
º º'-'
º
7\■ gº °ºº, L. J sº
sº sºvº º■C º
º
Sº

ºut■º
º

Sºmeº
~7.

C■º inci■co º
-

S
º %.
4.
º %.
sº c
sº-- *...* B RARY
S- *

sº Cº. Q-
1- º Z.

*-sº tºº,
*

...] º, O)le > •o LIBRARY & *

O) le sº
*
C■ S.
* [...] s
Qe

* […]
—r-

ºvº gº * %.--Sº
o ~ C ~ ºf ~

sºvº gº º*72,L. JS gº º(C

-

º sº
ºut■º
-- 'p, ºf º, sº cº

ºtcí■co
* , sº
*z, Nº wºoin' ///pSº / 42 Sº
sº Sº i■ºcºco %sº,
sº on
dº sº
%.S C■.
Cº
sº tº 22- -- tº tº 2- ºr- ºr
a

º 1- & º ~ º º o -

|s ºld "… *
ºvigº, º
'4. sº
º

TIC º sº'º ººvºgº - *

, sº
º cºnciº º, wº*/º º, Sºncisco … ºut º */ sº
sº
42
º
■ º

º,*o tº sº. Ole Hº tºº sº.

- ..Sº 't S ’9, *

O/] sº
- L_) º' º, L. e […] ‘o. C, &
ºg IT *, º, --sº
~ &
~/C º
[...] s ■ºvº. 9 in
º
*. 92 * - C
&
Oc C _º
-- sº
-
-

■º gº º put/? Sº
'a. Sº
C■.º/rancisco
º º >

r
&
42 Sº
sº C■º u■!Cº■co a' s %,
%.S.
A
M
&
*%.S.
*
%)
*

º
s *%,
*, 0.15}} *

[+º,
º º &

2– sº sº L. J º, O)le sº […] ”, Ll B| RA |R_Y sº L. º,

L! B RARY
- sº º, […] sº
-' sº ºvºi g : T %,”,L. sºJ s* ~ / o, .*
C ■ /C
*o * o
%2.*... sºsº ºvº g tº º, *... sº Sº %2.
~
sº ºut■º
12
º

º,
2 <
Sº 4. c) tº-

francisco º
> */
2. º
SºS 424. C■. 1 7///7. 7 //c■. To
C
z NS
ºut■º ■ * -- - --

ºº * O/2
[…] RA
[] sº s L_ º, Q)le sº […] *, LIBRARY s
*… […] sº * L. sººf
Y-sº- ”, “… 3.

| |B R_Y
S- 1- * °o. z

C vº■ °,º, sºsº º/C ”, º AQ: vº °. sº AT5 9 IT 9 : T.

ºncºco *z,sº.sº cºlour.*/7º 42sºSº C■º ºnci■co *2,gºNº º■ºSº
º
* *
* -

º/7 sº, &

42 Sº Cº. *
º

Sº
º

Yº t
º, S c

O/] sº. LIBRARY sº tº O/2- sº s […] º, Ll

-

S t- : Kº,

ARY & […]”.

º, -

- sº ~/C * [I] sº ºvº. 9 in * -- sº ~/C * [I] ■ºvº gº º

--- x O. * - O 9 Oc

º,
*~ C■.ºntº º,sº,sº d * ~~ FO º, sº a
R REFERENcé ■co sº,
º, sº Cºl/11////?'
*. º
- * > Lºº. tº 22- -
* g~,17 %, L. J sº -(■C º *** TAKEN FROM the soon § -(■C º%.[…] _S AR
AºS
A Sº ºf C■.ºutcº sº,
7/ N > º

ºncºco SS 4.
car

sº º a • * b. as a tz
– c. "
ºrs -
% - Cº.)
| |- sº tºº sº. $
º
*3 º
L gº %. ...)/ 2. y
... tº sº %.
tº ºLuº º
º

… […] s
> ~

T.º
O º, S

nº. Lºlº s “I■C

*
o

º,º,L. sºJ sº
ºf

º
. Sº
AQ#vº J
* * * * ** * * /2,~ º,
42 sº
Cy 7–
zº
42. SºSº
Jº■vº■ G 11
* **** * * * * * / , , , , , zºº
O
12, N o -
-(■C 7–
º,
- _
 -

|
-